Next Article in Journal
Correction: Hung et al. Cul4A Modulates Invasion and Metastasis of Lung Cancer through Regulation of ANXA10. Cancers 2019, 11, 618
Next Article in Special Issue
Histological Image Classification Between Follicular Lymphoma and Reactive Lymphoid Tissue Using Deep Learning and Explainable Artificial Intelligence (XAI)
Previous Article in Journal
Trends in Cancer Incidence and Associated Risk Factors in People Living with and Without HIV in Botswana: A Population-Based Cancer Registry Data Analysis from 1990 to 2021
Previous Article in Special Issue
Bridging the Gap Between Accuracy and Efficiency in AI-Based Breast Cancer Diagnosis from Histopathological Data
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Cancers
Volume 17
Issue 14
10.3390/cancers17142376
cancers-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Decoding the JAK-STAT Axis in Colorectal Cancer with AI-HOPE-JAK-STAT: A Conversational Artificial Intelligence Approach to Clinical–Genomic Integration

by
Ei-Wen Yang
1,
Brigette Waldrup
2 and
Enrique Velazquez-Villarreal
2,3,*
1
PolyAgent, San Francisco, CA 94102, USA
2
Department of Integrative Translational Sciences, Beckman Research Institute of City of Hope, Duarte, CA 91010, USA
3
City of Hope Comprehensive Cancer Center, Duarte, CA 91010, USA
*
Author to whom correspondence should be addressed.
Cancers 2025, 17(14), 2376; https://doi.org/10.3390/cancers17142376
Submission received: 8 June 2025 / Revised: 16 July 2025 / Accepted: 16 July 2025 / Published: 17 July 2025
(This article belongs to the Special Issue AI-Based Applications in Cancers)
Browse Figures
Review Reports Versions Notes

Simple Summary

Understanding how specific genetic changes affect colorectal cancer (CRC) is essential for improving treatment, especially in younger patients and those receiving chemotherapy. The JAK-STAT signaling pathway plays an important role in cancer development and immune response, but its impact on CRC has been difficult to study due to the complexity of available data. In this study, we developed AI-HOPE-JAK-STAT, a new artificial intelligence platform that allows researchers and clinicians to explore large cancer datasets simply by asking questions in plain English. This tool can quickly analyze how mutations in JAK-STAT genes relate to patient survival, treatment outcomes, and tumor characteristics. Our findings show that some JAK-STAT mutations are linked to better survival in younger patients and in those treated with common chemotherapy. By making complex data more accessible, AI-HOPE-JAK-STAT may accelerate cancer research and help guide more personalized treatment decisions in the future.

Abstract

Background/Objectives: The Janus kinase-signal transducer and activator of transcription (JAK-STAT) signaling pathway is a critical mediator of immune regulation, inflammation, and cancer progression. Although implicated in colorectal cancer (CRC) pathogenesis, its molecular heterogeneity and clinical significance remain insufficiently characterized—particularly within early-onset CRC (EOCRC) and across diverse treatment and demographic contexts. We present AI-HOPE-JAK-STAT, a novel conversational artificial intelligence platform built to enable the real-time, natural language-driven exploration of JAK/STAT pathway alterations in CRC. The platform integrates clinical, genomic, and treatment data to support dynamic, hypothesis-generating analyses for precision oncology. Methods: AI-HOPE-JAK-STAT combines large language models (LLMs), a natural language-to-code engine, and harmonized public CRC datasets from cBioPortal. Users define analytical queries in plain English, which are translated into executable code for cohort selection, survival analysis, odds ratio testing, and mutation profiling. To validate the platform, we replicated known associations involving JAK1, JAK3, and STAT3 mutations. Additional exploratory analyses examined age, treatment exposure, tumor stage, and anatomical site. Results: The platform recapitulated established trends, including improved survival among EOCRC patients with JAK/STAT pathway alterations. In FOLFOX-treated CRC cohorts, JAK/STAT-altered tumors were associated with significantly enhanced overall survival (p < 0.0001). Stratification by age revealed survival advantages in younger (age < 50) patients with JAK/STAT mutations (p = 0.0379). STAT5B mutations were enriched in colon adenocarcinoma and correlated with significantly more favorable trends (p = 0.0000). Conversely, JAK1 mutations in microsatellite-stable tumors did not affect survival, emphasizing the value of molecular context. Finally, JAK3-mutated tumors diagnosed at Stage I–III showed superior survival compared to Stage IV cases (p = 0.00001), reinforcing stage as a dominant clinical determinant. Conclusions: AI-HOPE-JAK-STAT establishes a new standard for pathway-level interrogation in CRC by empowering users to generate and test clinically meaningful hypotheses without coding expertise. This system enhances access to precision oncology analyses and supports the scalable, real-time discovery of survival trends, mutational associations, and treatment-response patterns across stratified patient cohorts.

1. Introduction

Colorectal cancer (CRC) remains one of the most prevalent and lethal malignancies worldwide, ranking third in global cancer incidence and second in cancer-related deaths [1,2,3]. While overall CRC rates have stabilized or declined in older adults, the incidence of early-onset colorectal cancer (EOCRC)—defined as diagnosis before the age of 50—continues to rise across several populations with disproportionate health burdens in the United States [4,5,6,7,8]. This increase has been accompanied by high rates of EOCRC-related mortality [9,10,11].
EOCRC exhibits distinct clinical and molecular features, including a higher frequency of microsatellite instability (MSI), increased tumor mutational burden, immune checkpoint activation, and unique epigenetic signatures [12,13,14,15,16]. These characteristics suggest alternative oncogenic mechanisms that may differ from those driving late-onset CRC (LOCRC), warranting further investigation into pathway-specific alterations across age and ancestry groups.
The Janus kinase-signal transducer and activator of transcription (JAK/STAT) pathway is a major signaling cascade involved in CRC pathogenesis. The JAK/STAT pathway remains underexplored in EOCRC. Aberrant JAK/STAT signaling has been associated with chronic inflammation, tumor immune evasion, and epithelial–mesenchymal transition (EMT), contributing to more aggressive tumor behavior [1,17,18,19,20,21,22] and reduced treatment response in CRC [23,24,25,26,27]. STAT3 activation, in particular, has been linked to stromal invasion and adverse survival outcomes [28,29,30].
Recent studies investigating molecular alterations in JAK/STAT pathway genes among CRC patients have reported higher frequencies of JAK/STAT-related mutations in EOCRC cases among individuals [20,31,32] from populations with different ethnicities [1]. However, no significant differences in JAK/STAT pathway alterations were observed across age or ancestry groups. Survival analyses indicated that JAK/STAT alterations may still be associated with clinical outcomes, although findings were variable across subgroups. These observations highlight the need for tools that can enable the detailed, flexible exploration of molecular data across clinical and genetic contexts.
Advances in artificial intelligence (AI), including the application of large language models (LLMs), are transforming our ability to extract insights from complex biomedical datasets [33,34,35,36,37]. Conversational AI systems can interpret natural language queries and execute analytical workflows in real time, making it easier for researchers and clinicians to interact with genomic and clinical data [38,39,40]. Despite their promise, platforms focused on specific pathway-level interrogation are essential in enabling accurate analysis and integration across multiple data modalities.
To meet this need, we developed AI-HOPE-JAK-STAT, a conversational AI agent specifically designed to investigate JAK/STAT pathway alterations in CRC. The system enables the integration of clinical, genomic, and transcriptomic data through natural language-driven analysis. In this study, we describe the development of AI-HOPE-JAK-STAT, demonstrate its analytical performance by replicating known trends in CRC cohorts, and explore its application in identifying clinically relevant patterns in EOCRC across population groups.
AI-HOPE-JAK-STAT extends our growing suite of conversational AI platforms designed for precision oncology, building on the foundational architectures of AI-HOPE [38], AI-HOPE-TGFbeta [39], and AI-HOPE-PI3K [40]. While previous agents focused on general clinical–genomic queries or specific pathways such as TGF-β and PI3K, AI-HOPE-JAK-STAT uniquely targets the JAK/STAT signaling axis—a critical, yet underexplored, pathway in colorectal cancer. This platform incorporates lessons learned from our prior systems, particularly in optimizing prompt engineering, enhancing reproducibility, and dynamically tailoring analytical workflows in response to user input. Like its predecessors, AI-HOPE-JAK-STAT supports key statistical operations including survival analysis, mutation frequency comparisons, and odds ratio estimation; however, it is distinguished by its ability to stratify analyses by variables such as treatment exposure, microsatellite status, and ancestry. By embedding these functionalities into a natural language-driven interface, AI-HOPE-JAK-STAT empowers users to interrogate the clinical significance of JAK/STAT alterations in real time, facilitating both exploratory hypothesis generation and pathway-specific biomarker discovery.

2. Materials and Methods

2.1. AI-HOPE-JAK-STAT Platform Design and Functionality

AI-HOPE-JAK-STAT is a custom-built conversational artificial intelligence platform designed to explore the clinical and genomic landscape of JAK/STAT pathway alterations in CRC. The system leverages a LLM-driven interface that translates plain language queries into executable bioinformatics commands (Figure 1). This architecture enables the real-time integration and interrogation of multi-dimensional CRC datasets using a modular backend combining data parsing, statistical modeling, and dynamic visualization.
To ensure reproducibility and accessibility, we have made the full AI-HOPE-JAK-STAT codebase, including natural language-to-code translation logic, data processing modules, and backend bioinformatics workflows, publicly available (see data accessibility statement). The platform uses a LLaMA 3-based large language model deployed locally with predefined prompts optimized for clinical–genomic query translation. It operates via a Python v3.12-based interpreter that converts natural language inputs into executable code, leveraging commonly used packages (e.g., pandas, lifelines, matplotlib) for real-time survival analysis, cohort filtering, and mutation profiling. The system is fully open-source and designed for research use only. This public release ensures that others can reproduce the results and extend the platform for additional use cases in colorectal cancer and beyond.
To provide greater clarity on the system’s technical implementation, AI-HOPE-JAK-STAT is built on a modular Python framework that integrates a locally deployed LLaMA 3-based large language model. The model processes user queries via custom-designed prompts that have been fine-tuned to optimize accuracy in clinical–genomic query translation. The backend uses well-established Python libraries—including pandas for data manipulation, lifelines for survival analysis, matplotlib for visualization, and scipy for statistical testing—to generate and execute code in real time. Each plain language query is parsed through a structured prompt pipeline that extracts key parameters (e.g., gene, age group, treatment exposure) and dynamically constructs the appropriate statistical workflow. The system operates locally for full data privacy and computational control, and its codebase—including prompt logic, data ingestion scripts, statistical modules, and visualization functions—is publicly available (see the data availability statement). This ensures full transparency, reproducibility, and adaptability for users seeking to apply or extend the platform to new datasets or cancer types.

2.2. Data Sources and Processing

The platform ingests harmonized CRC datasets derived from the publicly available repository cBioPortal. Data preprocessing involved the normalization of sample identifiers, the removal of incomplete clinical records, and the mapping of genomic features to canonical gene symbols based on HGNC nomenclature. Genes involved in the JAK/STAT signaling pathway—such as JAK1, JAK2, JAK3, TYK2, STAT1, STAT3, STAT5A/B, and SOCS family members—were predefined for targeted analysis. Clinical annotations included tumor stage, anatomical location, treatment history, microsatellite instability (MSI) status, and self-reported ancestry.

2.3. Natural Language Input and Query Interpretation

Users interact with the platform through an intuitive natural language interface. AI-HOPE-JAK-STAT processes these inputs using an LLM (LLaMA 3-based) to identify intent, extract parameters, and structure downstream analytic operations. Example queries include the following: “Compare mutation rates in STAT3 between patients under and over 50,” or “Show survival differences in JAK1-mutated tumors treated with chemotherapy.” Ambiguities in user input trigger clarification prompts to maintain analytic precision and reproducibility.
To further clarify the LLM component of AI-HOPE-JAK-STAT, the platform employs a locally deployed LLaMA 3-based model configured with inference parameters optimized for reproducibility and precision: temperature = 0.2, top_p = 0.9, and a maximum token length of 2048. These settings were selected to reduce response variability and ensure the consistent translation of natural language inputs into executable code. The system uses structured, domain-specific prompt templates designed to extract analytical intent from user queries. Each prompt includes slots for gene targets, clinical stratifiers (e.g., age, treatment, MSI status), and desired outputs (e.g., Kaplan–Meier plots, contingency tables). The model then translates these prompts into code using predefined syntax rules. Importantly, AI-HOPE-JAK-STAT runs entirely locally, without reliance on third-party APIs, ensuring user data security and analytical reproducibility. If a different LLM is adopted in future versions, the prompt templates may require re-tuning to ensure continued accuracy and relevance. This flexibility has been built into the platform’s modular design to allow for future upgrades.

2.4. Analytical Framework

Backend analytics for AI-HOPE-JAK-STAT are implemented in Python, leveraging libraries such as pandas, lifelines, and SciPy v1.11 to execute a range of statistical and bioinformatics operations. The platform performs descriptive summaries to assess mutation frequencies and characterize cohort composition. For categorical comparisons, it employs chi-square or Fisher’s exact tests to evaluate statistical significance. Survival outcomes are analyzed using Kaplan–Meier estimates with log-rank testing, while multivariable survival modeling is conducted using Cox proportional hazards regression. The system also enables flexible subgroup stratification based on age group, self-reported ancestry, treatment exposure, and tumor location. Additionally, AI-HOPE-JAK-STAT supports co-occurrence and mutual exclusivity analyses to explore potential interactions between JAK/STAT pathway genes and other molecular drivers relevant to colorectal tumorigenesis.

2.5. Application Validation

To evaluate AI-HOPE-JAK-STAT’s analytical robustness, we applied the platform to replicate findings from previous large-cohort studies investigating JAK/STAT and MAPK pathway alterations across EOCRC and LOCRC populations. Validation tasks included mutation frequency comparisons between different ethnicity EOCRC patients and assessment of survival impact among STAT3-mutated cases.

2.6. Output Generation and Interpretation

Following analysis, the system generates comprehensive visual and tabular outputs. These include survival curves, mutation distribution plots, heatmaps, and odds ratio forest plots. Results are accompanied by auto-generated narrative summaries linking observed patterns to published biomedical studies through a retrieval-augmented generation (RAG) module. All outputs are exportable for downstream reporting or figure preparation.
To evaluate the reproducibility of AI-HOPE-JAK-STAT outputs, we conducted repeated analyses using identical natural language queries across multiple runs and computational environments. We confirmed that survival curves, statistical outputs (e.g., p-values, odds ratios), and visualizations remained consistent, demonstrating the platform’s reliability in producing stable results. Reproducibility is further supported by version-controlled backend scripts and automated query logging, allowing users to trace and replicate analytical workflows. To enhance the interpretability of the results, the platform also includes a Retrieval-Augmented Generation (RAG) module that generates narrative summaries linking observed findings to relevant biomedical studies. This module retrieves content from a curated index of open-access sources, including selected journals with permissive licensing. Importantly, all RAG-generated summaries presented in this study were manually reviewed and curated by scientific domain experts to ensure contextual relevance, scientific accuracy, and alignment with the underlying data.

2.7. Usability and Comparative Evaluation

To assess platform usability, we benchmarked AI-HOPE-JAK-STAT against existing tools (e.g., cBioPortal, Xena Browser) across common analytic tasks such as stratified mutation analysis, co-alteration assessment, and treatment–outcome comparisons. The metrics evaluated included the task completion time, flexibility in cohort definition, and reproducibility of results. AI-HOPE-JAK-STAT demonstrated increased efficiency, especially in generating nested subgroup analyses and complex multi-parameter queries.

3. Results

AI-HOPE-JAK-STAT enabled the natural language-driven interrogation of the JAK/STAT signaling cascade across diverse CRC patient populations, integrating genomic alterations with clinical outcomes in real time. Through a series of validation and exploratory analyses, the platform demonstrated its ability to reproduce known associations, uncover novel patterns, and generate interpretable visual and statistical outputs that facilitate hypothesis generation and precision oncology discovery.
To assess ancestry-specific differences in JAK/STAT pathway alterations and their potential clinical significance, we conducted focused analyses using AI-HOPE-JAK-STAT in early-onset colorectal cancer (EOCRC) cohorts stratified by ancestry. As a first example, we queried Hispanic/Latino (H/L) EOCRC patients under 50 years of age and compared overall survival between those harboring JAK/STAT pathway mutations (n = 16) and those without such alterations (n = 137). In this analysis, presented in Figure 2, the observed p-value of 0.0539 does not meet the conventional threshold for statistical significance. Accordingly, we interpret this finding as non-significant and hypothesis-generating, rather than indicative of a definitive survival benefit. While the trend may suggest a potential prognostic signal, it should be interpreted with caution and warrants further validation in larger, independent H/L-specific CRC cohorts. To evaluate the robustness and ancestry-specific reproducibility of this observation, we replicated the analysis in EOCRC patients of Non-Hispanic White (NHW) descent. In this group, patients with JAK/STAT pathway alterations demonstrated a statistically significant survival advantage compared to those without (p = 0.0001; Figure S1), reinforcing the potential prognostic value of these mutations. These ancestry-stratified examples illustrate how AI-HOPE-JAK-STAT enables the real-time, hypothesis-driven interrogation of clinical and genomic relationships that would otherwise require extensive manual coding and data wrangling. Although the current version of the tool is designed for research use only, its ability to rapidly contextualize genomic alterations within demographic and clinical subgroups provides a scalable framework for future applications in biomarker discovery and precision oncology.
Next, we investigated treatment context by focusing on CRC patients who received FOLFOX chemotherapy. Kaplan–Meier survival analysis revealed that individuals with JAK/STAT pathway alterations (n = 282) had significantly better survival compared to those without such mutations (n = 3791) under the same treatment regimen (p < 0.0001; Figure 3). These findings suggest that JAK/STAT alterations may be associated with enhanced treatment response in FOLFOX-exposed patients.
To explore age-related effects, we stratified JAK/STAT-altered CRC patients by early- (<50 years; n = 133) versus late-onset (≥50 years; n = 424) status. Early-onset cases exhibited significantly improved survival (p = 0.0379), and odds ratio testing suggested modest enrichment of FOLFOX treatment in this younger group, although this was not statistically significant (OR = 1.356, p = 0.154; Figure 4). These results highlight possible interactions between age, treatment exposure, and molecular alterations.
We then assessed the prognostic impact of JAK1 mutations in microsatellite-stable (MSS) tumors. Among MSS CRC patients, those with JAK1 mutations (n = 44) showed no significant difference in overall survival relative to wild-type controls (n = 4638; p = 0.5691; Figure 5). Despite the absence of survival benefit, this analysis underscores the utility of AI-HOPE-JAK-STAT in isolating molecular contexts for precision stratification.
Further, we analyzed STAT5B mutation status in primary tumor samples. Patients with STAT5B-mutated primary tumors (n = 109) experienced significantly better overall survival than those without such alterations (n = 3945; p < 0.0001; as shown in Figure S2). Odds ratio testing revealed that STAT5B-mutated tumors were more frequently located in the colon than in other sites (OR = 1.949, p = 0.002), suggesting anatomical preferences in mutation distribution and potential therapeutic relevance.
Finally, we explored stage-specific survival patterns among JAK3-mutated CRC patients. When stratified by disease stage, individuals diagnosed at Stage I–III (n = 142) exhibited significantly more favorable survival trends than those with Stage IV tumors (n = 36), with highly significant survival differences (p < 0.0000; Figure S3). These findings underscore the strong prognostic impact of clinical stage even within molecularly defined subgroups.
Together, these analyses establish AI-HOPE-JAK-STAT as a robust, interactive platform capable of uncovering clinically relevant associations between JAK/STAT signaling alterations and survival outcomes across stratified CRC cohorts. By combining natural language interfaces with integrated bioinformatics, the system supports the reproducible, real-time analysis of complex cancer datasets, offering new pathways toward individualized treatment strategies.

4. Discussion

This study presents AI-HOPE-JAK-STAT, a novel conversational artificial intelligence platform designed to interrogate the clinical–genomic landscape of JAK/STAT signaling alterations in CRC. By combining a natural language interface with automated, reproducible bioinformatics workflows, the system addresses longstanding limitations in accessibility and flexibility of pathway-specific data analysis, particularly for EOCRC and other clinically stratified subgroups.
Unlike conventional tools that rely on manual curation or scripting expertise, AI-HOPE-JAK-STAT enables users to conduct complex, multi-layered analyses through natural language queries. Across validation and exploratory tasks, the platform replicated known associations—such as the prognostic impact of JAK/STAT alterations in EOCRC—and uncovered novel trends in treatment response and mutational co-enrichment. In doing so, AI-HOPE-JAK-STAT has demonstrated that conversational AI can serve as a scalable and robust solution for precision oncology research.
Our findings reinforce the biological and clinical relevance of JAK/STAT signaling in CRC. In early-onset patients of both Hispanic/Latino (H/L) and NHW backgrounds, alterations in pathway genes such as JAK1, JAK3, and STAT5B were associated with improved survival, particularly among those receiving FOLFOX chemotherapy. These results align with the preclinical literature suggesting that JAK/STAT dysregulation may sensitize tumors to immune-mediated or cytotoxic interventions. Moreover, the observed stage-specific survival differences among JAK3-mutant tumors (Stage I–III vs. Stage IV) suggest that clinical context significantly modulates the prognostic impact of these alterations.
Beyond outcome comparisons, AI-HOPE-JAK-STAT was also effective in evaluating population-level enrichment. For example, STAT5B mutations were found to be significantly more prevalent in colon versus rectal tumors, an anatomical preference that may reflect divergent etiologies or tissue-specific gene regulation. Conversely, JAK1 mutations within microsatellite-stable (MSS) tumors did not exhibit a survival effect, emphasizing the importance of molecular context in interpreting prognostic significance. Such nuanced insights are rarely obtainable without substantial analytic infrastructure, yet were generated in seconds using AI-HOPE-JAK-STAT’s conversational engine.
A key strength of the platform lies in its adaptability across user-defined comparisons—age, ancestry, tumor stage, and treatment regimen—all seamlessly integrated through structured outputs. This flexibility positions AI-HOPE-JAK-STAT as not just an analytics tool, but a collaborative system for hypothesis testing, education, and translational discovery. The integration of retrieval-augmented generation (RAG) further enhances interpretability by contextualizing results within the existing literature, supporting evidence-based insight generation.
We recognize the importance of reproducibility and contextual reliability in AI-assisted research platforms. In this revision, we strengthened the reproducibility claims of AI-HOPE-JAK-STAT by performing validation tests across multiple executions of the same query, confirming consistency in statistical outputs and visualizations. Moreover, we clarified the role and limitations of the RAG module. While this feature enriches outputs by surfacing the relevant literature, it draws from a curated selection of open-access databases and may not reflect the full breadth of biomedical publications due to indexing or licensing restrictions. As such, the RAG-generated summaries should be interpreted as exploratory aids rather than definitive clinical annotations. These revisions underscore our commitment to transparency, reliability, and the responsible use of AI tools in precision oncology.
We acknowledge that several subgroup analyses presented in this study involve limited sample sizes, which constrain the statistical power and reliability of certain findings. These analyses—such as the 16 early-onset H/L cases in Figure 2 and the 44 JAK1-mutated MSI-Stable patients in Figure 5—are explicitly exploratory in nature and should be interpreted as hypothesis-generating rather than confirmatory. AI-HOPE-JAK-STAT operates as an intelligent conversational system whose analytical capabilities are built upon statistical frameworks validated in our previous work. These prior studies guided the system’s training, the design of prompt-to-code translation, and the integration of standardized statistical methods (e.g., survival analysis via Kaplan–Meier estimates and log-rank testing). Because the platform leverages harmonized, publicly available data from cBioPortal, subgroup availability is inherently constrained by the underlying datasets. Importantly, generating accurate outputs from small, stratified cohorts requires iterative optimization within the system’s inference logic. As such, we emphasize that the analyses presented here are meant to demonstrate the platform’s exploratory and interactive capabilities, and advise against drawing definitive clinical conclusions.
In our analysis of JAK3-mutated colorectal tumors, we observed significantly improved survival among patients diagnosed at Stage I–III compared to those at Stage IV (p < 0.00001; Figure S3). While this observation demonstrates the system’s ability to stratify survival outcomes by both mutation status and clinical variables, we acknowledge that the survival difference is most likely driven by stage rather than JAK3 mutation itself serving as an independent prognostic factor. This highlights the importance of incorporating multivariable adjustment to account for confounding variables such as tumor stage, treatment exposure, and MSI status. Future versions of AI-HOPE-JAK-STAT will include built-in support for automated Cox proportional hazards modeling to enable more rigorous, adjusted survival analyses, enhancing the platform’s ability to deliver clinically meaningful and context-aware insights.
The current version of AI-HOPE-JAK-STAT was originally configured and optimized based on the analytical framework used in our prior publication [1], which served as a reference standard for training and validating the system’s performance. The platform integrates predefined statistical logic, subgroup structures, and survival analysis pipelines derived from that work, enabling the dynamic execution of queries using harmonized colorectal cancer datasets from cBioPortal. These foundations ensure that the system generates reproducible and interpretable outputs for hypothesis-driven research. However, while the existing version performs univariate survival analysis (e.g., Kaplan–Meier estimates with log-rank tests), it was not initially configured to automate multivariable modeling. Future developments of AI-HOPE-JAK-STAT will incorporate built-in support for multivariable survival modeling and statistical power estimation, further enhancing the platform’s capability to support robust, exploratory, and translational analyses in precision oncology.
AI-HOPE-JAK-STAT represents the next evolution in our series of conversational AI platforms for precision oncology, offering a pathway-specific framework tailored to interrogate the clinical relevance of JAK/STAT signaling in colorectal cancer. Compared to our earlier agents—AI-HOPE [39], which established the feasibility of natural language-guided multi-omics exploration, and AI-HOPE-TGFbeta [40] and AI-HOPE-PI3K [41], which focused on discrete oncogenic pathways—AI-HOPE-JAK-STAT introduces refined capabilities for handling complex stratification variables such as treatment history, microsatellite status, and self-reported ancestry. This platform also integrates an updated prompt engine, expanded backend analytics, and a reproducibility-focused infrastructure that builds on insights gained from prior iterations. Notably, JAK/STAT pathway alterations present unique analytical challenges due to their diverse roles in immune modulation and tumor progression, and AI-HOPE-JAK-STAT is specifically designed to accommodate these complexities. Together, these advancements demonstrate the growing versatility of our AI-HOPE framework and its potential to support hypothesis-driven, equitable research across biologically distinct signaling pathways in colorectal cancer.
Nevertheless, several limitations warrant discussion. The reliance on publicly available datasets, such as those from cBioPortal, may limit the generalizability of findings due to sample size constraints and potential biases in data curation, particularly the under-representation of certain ancestry groups and uneven clinical annotation across studies. Although AI-HOPE-JAK-STAT supports stratification by ancestry, tumor stage, and treatment, its outputs are inherently shaped by the composition and completeness of the source data. Broader deployment and more equitable insight generation will benefit from the integration of additional multi-omic layers (e.g., RNA-seq [10,41], spatial biology data [42,43,44] and the inclusion of prospective, demographically diverse clinical datasets. Furthermore, while the natural language interface promotes accessibility, performance can vary depending on query specificity. Future iterations may incorporate query refinement tools, dialogue memory, and user feedback loops to enhance interpretability, robustness, and overall user experience.
Despite these limitations, the platform’s performance in recapitulating prior findings, identifying novel associations, and supporting pathway-level analyses without coding expertise reflects a meaningful advancement in AI-enabled cancer research. AI-HOPE-JAK-STAT not only complements traditional tools like cBioPortal and Xena but offers unique advantages in speed, flexibility, and hypothesis exploration.

5. Conclusions

AI-HOPE-JAK-STAT offers a promising addition to the evolving precision oncology landscape by enabling the natural language-guided integration of genomic, clinical, and treatment data for the investigation of JAK/STAT pathway alterations in colorectal cancer (CRC). The platform facilitates real-time, hypothesis-driven analyses—such as survival trends, mutation enrichment, and ancestry- or treatment-specific subgroup exploration—without requiring programming expertise. These features may be particularly useful in examining the heterogeneity of early-onset CRC and generating insights that could support the design of future clinical studies and personalized treatment strategies. While AI-HOPE-JAK-STAT demonstrates the feasibility of conversational AI for pathway-level analysis, its broader utility will depend on continued development, the integration of multivariable modeling, and validation across more diverse, prospective datasets. As precision oncology moves toward more individualized and immunologically informed care, tools like AI-HOPE-JAK-STAT may help support biomarker discovery and enhance accessibility to integrative data analysis across varied research and clinical environments.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cancers17142376/s1, Figure S1: AI-HOPE-JAK-STAT analysis of early-onset colorectal cancer in patients stratified by JAK/STAT pathway alteration status. Figure S2: AI-HOPE-JAK-STAT analysis of STAT5B-mutant primary colorectal tumors with odds ratio context in colon adenocarcinoma. Figure S3: AI-HOPE-JAK-STAT analysis of colorectal cancer patients with JAK3 mutations stratified by tumor stage.

Author Contributions

Conceptualization, E.V.-V., E.-W.Y. and B.W.; methodology, E.V.-V. and B.W.; software, E.V.-V. and E.-W.Y.; validation, B.W., E.-W.Y. and E.V.-V.; formal analysis, E.V.-V. and B.W.; investigation, E.V.-V. and B.W.; resources, E.V.-V.; data curation, B.W. and E.V.-V.; writing—original draft preparation, E.V.-V. and E.-W.Y.; writing—review and editing, E.V.-V., B.W. and E.-W.Y.; visualization, E.V.-V. and B.W.; supervision, E.V.-V.; project administration, E.V.-V.; funding acquisition, E.V.-V. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Department of Integrative Translational Sciences at City of Hope, the Cancer Control and Population Sciences Program at City of Hope, and the National Institutes of Health (NIH) and the National Cancer Institute (NCI) through the following awards: P30-CA033572; the Cancer Moonshot PE-CGS project titled “Optimizing Engagement of Hispanic Colorectal Cancer Patients in Cancer Genomic Characterization Studies” (U2C-CA252971); and the U54 partnership between the University of California, Riverside (UCR) and the City of Hope Comprehensive Cancer Center (COH-CCC) focused on drug development and capacity building (U54-CA285116).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data used in this study is available to the public and can be found at cbioportal.org. The AI-HOPE-JAK-STAT software, along with demonstration data and documentation, is available at https://github.com/Velazquez-Villarreal-Lab/AI-JAK-STAT (accessed on 11 July 2025).

Acknowledgments

The authors wish to thank the Department of Integrative Translational Sciences at City of Hope, the Cancer Control and Population Sciences Program at the City of Hope Comprehensive Cancer Center (NIH P30 CA033572), and the Drug Development and Capacity Building: A UCR/CoH-CCC Partnership initiative (NIH U54 CA285116) for their support.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Monge, C.; Waldrup, B.; Carranza, F.G.; Velazquez-Villarreal, E. Ethnicity-Specific Molecular Alterations in MAPK and JAK/STAT Pathways in Early-Onset Colorectal Cancer. Cancers 2025, 17, 1093. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  2. Sung, H.; Ferlay, J.; Siegel, R.L.; Laversanne, M.; Soerjomataram, I.; Jemal, A.; Bray, F. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J. Clin. 2021, 71, 209–249. [Google Scholar] [CrossRef] [PubMed]
  3. Bhandari, A.; Woodhouse, M.; Gupta, S. Colorectal cancer is a leading cause of cancer incidence and mortality among adults younger than 50 years in the USA: A SEER-based analysis with comparison to other young-onset cancers. J. Investig. Med. 2017, 65, 311–315. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  4. Patel, S.G.; Karlitz, J.J.; Yen, T.; Lieu, C.H.; Boland, C.R. The rising tide of early-onset colorectal cancer: A comprehensive review of epidemiology, clinical features, biology, risk factors, prevention, and early detection. Lancet. Gastroenterol. Hepatol. 2022, 7, 262–274. [Google Scholar] [CrossRef] [PubMed]
  5. Santucci, C.; Mignozzi, S.; Alicandro, G.; Pizzato, M.; Malvezzi, M.; Negri, E.; Jha, P.; La Vecchia, C. Trends in cancer mortality under age 50 in 15 upper-middle and high-income countries. J. Natl. Cancer Inst. 2025, 117, 747–760. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  6. Sinicrope, F.A. Increasing Incidence of Early-Onset Colorectal Cancer. N. Engl. J. Med. 2022, 386, 1547–1558. [Google Scholar] [CrossRef] [PubMed]
  7. Garcia, S.; Pruitt, S.L.; Singal, A.G.; Murphy, C.C. Colorectal cancer incidence among Hispanics and non-Hispanic Whites in the United States. Cancer Causes Control 2018, 29, 1039–1046. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  8. Monge, C.; Waldrup, B.; Manjarrez, S.; Carranza, F.G.; Velazquez-Villarreal, E. Detecting PI3K and TP53 Pathway Disruptions in Early-Onset Colorectal Cancer Among Hispanic/Latino Patients. Cancer Med. 2025, 14, e70791. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  9. Monge, C.; Waldrup, B.; Carranza, F.G.; Velazquez-Villarreal, E. WNT and TGF-Beta Pathway Alterations in Early-Onset Colorectal Cancer Among Hispanic/Latino Populations. Cancers 2024, 16, 3903. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  10. Müller, M.F.; Ibrahim, A.E.; Arends, M.J. Molecular pathological classification of colorectal cancer. Virchows Arch. 2016, 469, 125–134. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  11. Venugopal, A.; Carethers, J.M. Epidemiology and biology of early onset colorectal cancer. EXCLI J. 2022, 21, 162–182. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  12. Torres, T.; Arellano Villanueva, E.; Alsabawi, Y.; Fofana, D.; Tripathi, M.K. Unraveling Early Onset Disparities and Determinants: An Analysis of Colorectal Cancer Outcomes and Trends in Texas. Cureus 2025, 17, e83124. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  13. Kim, J.H.; Park, H.E.; Cho, N.Y.; Lee, H.S.; Kang, G.H. Characterisation of PD-L1-positive subsets of microsatellite-unstable colorectal cancers. Br. J. Cancer. 2016, 115, 490–496. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  14. Lu, C.; Zhang, X.; Schardey, J.; Wirth, U.; Heinrich, K.; Massiminio, L.; Cavestro, G.M.; Neumann, J.; Bazhin, A.V.; Werner, J.; et al. Molecular characteristics of microsatellite stable early-onset colorectal cancer as predictors of prognosis and immunotherapeutic response. npj Precis. Oncol. 2023, 7, 63. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  15. Mauri, G.; Sartore-Bianchi, A.; Russo, A.G.; Marsoni, S.; Bardelli, A.; Siena, S. Early-onset colorectal cancer in young individuals. Mol. Oncol. 2019, 13, 109–131. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  16. Lieu, C.H.; Golemis, E.A.; Serebriiskii, I.G.; Newberg, J.; Hemmerich, A.; Connelly, C.; Messersmith, W.A.; Eng, C.; Eckhardt, S.G.; Frampton, G.; et al. Comprehensive Genomic Landscapes in Early and Later Onset Colorectal Cancer. Clin. Cancer Res. 2019, 25, 5852–5858. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  17. Antelo, M.; Balaguer, F.; Shia, J.; Shen, Y.; Hur, K.; Moreira, L.; Cuatrecasas, M.; Bujanda, L.; Giraldez, M.D.; Takahashi, M.; et al. A high degree of LINE-1 hypomethylation is a unique feature of early-onset colorectal cancer. PLoS ONE 2012, 7, e45357. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  18. Lim, H.I.; Kim, G.Y.; Choi, Y.J.; Lee, K.; Ko, S.G. Uncovering the anti-cancer mechanism of cucurbitacin D against colorectal cancer through network pharmacology and molecular docking. Discov. Oncol. 2025, 16, 551. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  19. Xu, C.; Zhu, R.; Dai, Q.; Li, Y.; Hu, G.; Tao, K.; Xu, Y.; Xu, G.; Zhang, G. TIMP-2 Modulates 5-Fu Resistance in Colorectal Cancer Through Regulating JAK-STAT Signalling Pathway. J. Cell. Mol. Med. 2025, 29, e70470. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  20. Liu, J.; Han, S.; Cui, Y.; Zhao, Q.; Wang, Y.; Li, T.; Li, X. LncRNA FAM30A Suppresses Proliferation and Metastasis of Colorectal Carcinoma by Blocking the JAK-STAT Signalling. J. Cell. Mol. Med. 2025, 29, e70421. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  21. Johnson, D.E.; O’Keefe, R.A.; Grandis, J.R. Targeting the IL-6/JAK/STAT3 signalling axis in cancer. Nat. Rev. Clin. Oncol. 2018, 15, 234–248. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  22. Xue, J.; Liao, L.; Yin, F.; Kuang, H.; Zhou, X.; Wang, Y. LncRNA AB073614 induces epithelial- mesenchymal transition of colorectal cancer cells via regulating the JAK/STAT3 pathway. Cancer Biomark. 2018, 21, 849–858, Erratum in: Cancer Biomark. 2021, 32, 569. https://doi.org/10.3233/CBM-219780. [Google Scholar] [CrossRef] [PubMed]
  23. Villarino, A.V.; Kanno, Y.; Ferdinand, J.R.; O’Shea, J.J. Mechanisms of Jak/STAT signaling in immunity and disease. J. Immunol. 2015, 194, 21–27. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  24. Nobel, Y.R.; Stier, K.; Krishnareddy, S. STAT signaling in the intestine. Int. Rev. Cell Mol. Biol. 2021, 361, 1–20. [Google Scholar] [CrossRef] [PubMed]
  25. Awasthi, N.; Liongue, C.; Ward, A.C. STAT proteins: A kaleidoscope of canonical and non-canonical functions in immunity and cancer. J. Hematol. Oncol. 2021, 14, 198. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  26. Sabaawy, H.E.; Ryan, B.M.; Khiabanian, H.; Pine, S.R. JAK/STAT of all trades: Linking inflammation with cancer development, tumor progression and therapy resistance. Carcinogenesis 2021, 42, 1411–1419. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  27. Farooqi, A.A.; de la Roche, M.; Djamgoz, M.B.A.; Siddik, Z.H. Overview of the oncogenic signaling pathways in colorectal cancer: Mechanistic insights. Semin. Cancer Biol. 2019, 58, 65–79. [Google Scholar] [CrossRef] [PubMed]
  28. de Jong, P.R.; Mo, J.H.; Harris, A.R.; Lee, J.; Raz, E. STAT3: An Anti-Invasive Factor in Colorectal Cancer? Cancers 2014, 6, 1394–1407. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  29. Liu, H.; Ren, G.; Wang, T.; Chen, Y.; Gong, C.; Bai, Y.; Wang, B.; Qi, H.; Shen, J.; Zhu, L.; et al. Aberrantly expressed Fra-1 by IL-6/STAT3 transactivation promotes colorectal cancer aggressiveness through epithelial-mesenchymal transition. Carcinogenesis 2015, 36, 459–468. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  30. Pennel, K.A.F.; Hatthakarnkul, P.; Wood, C.S.; Lian, G.Y.; Al-Badran, S.S.F.; Quinn, J.A.; Legrini, A.; Inthagard, J.; Alexander, P.G.; van Wyk, H.; et al. JAK/STAT3 represents a therapeutic target for colorectal cancer patients with stromal-rich tumors. J. Exp. Clin. Cancer Res. 2024, 43, 64. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  31. Zhang, Z.; Wang, A.; Tang, X.; Chen, Y.; Tang, E.; Jiang, H. Comparative mutational analysis of distal colon cancer with rectal cancer. Oncol. Lett. 2020, 19, 1781–1788. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  32. Zhang, N.; Yue, W.; Jiao, B.; Cheng, D.; Wang, J.; Liang, F.; Wang, Y.; Liang, X.; Li, K.; Liu, J.; et al. Unveiling prognostic value of JAK/STAT signaling pathway related genes in colorectal cancer: A study of Mendelian randomization analysis. Infect. Agents Cancer 2025, 20, 9. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  33. Tordjman, M.; Bolger, I.; Yuce, M.; Restrepo, F.; Liu, Z.; Dercle, L.; McGale, J.; Meribout, A.L.; Liu, M.M.; Beddok, A.; et al. Large Language Models in Cancer Imaging: Applications and Future Perspectives. J. Clin. Med. 2025, 14, 3285. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  34. Shanwetter Levit, N.; Saban, M. When investigator meets large language models: A qualitative analysis of cancer patient decision-making journeys. npj Digit. Med. 2025, 8, 336. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  35. Buvat, I.; Weber, W. Nuclear Medicine from a Novel Perspective: Buvat and Weber Talk with OpenAI’s ChatGPT. J. Nucl. Med. 2023, 64, 505–507. [Google Scholar] [CrossRef] [PubMed]
  36. Visvikis, D.; Lambin, P.; Beuschau Mauridsen, K.; Hustinx, R.; Lassmann, M.; Rischpler, C.; Shi, K.; Pruim, J. Application of artificial intelligence in nuclear medicine and molecular imaging: A review of current status and future perspectives for clinical translation. Eur. J. Nucl. Med. Mol. Imaging 2022, 49, 4452–4463. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  37. Xu, J.; Yang, P.; Xue, S.; Sharma, B.; Sanchez-Martin, M.; Wang, F.; Beaty, K.A.; Dehan, E.; Parikh, B. Translating cancer genomics into precision medicine with artificial intelligence: Applications, challenges and future perspectives. Hum. Genet. 2019, 138, 109–124. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  38. Yang, E.W.; Velazquez-Villarreal, E. AI-HOPE: An AI-driven conversational agent for enhanced clinical and genomic data integration in precision medicine research. Bioinformatics 2025, 41, btaf359. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  39. Yang, E.-W.; Waldrup, B.; Velazquez-Villarreal, E. AI-HOPE-TGFbeta: A Conversational AI Agent for Integrative Clinical and Genomic Analysis of TGF-β Pathway Alterations in Colorectal Cancer to Advance Precision Medicine. AI 2025, 6, 137. [Google Scholar] [CrossRef]
  40. Yang, E.-W.; Waldrup, B.; Velazquez-Villarreal, E. From Mutation to Prognosis: AI-HOPE-PI3K Enables Artificial Intelligence Agent-Driven Integration of PI3K Pathway Data in Colorectal Cancer Precision Medicine. Int. J. Mol. Sci. 2025, 26, 6487. [Google Scholar] [CrossRef] [PubMed]
  41. Li, J.; Pan, Y.; Guo, F.; Wang, C.; Liang, L.; Li, P.; Liang, W.; Lian, P.; Chen, Y.; Yang, Y.; et al. Patterns in genomic mutations among patients with early-onset colorectal cancer: An international, multicohort, observational study. Lancet Oncol. 2025. Epub ahead of print. [Google Scholar] [CrossRef] [PubMed]
  42. Carranza, F.G.; Diaz, F.C.; Ninova, M.; Velazquez-Villarreal, E. Current state and future prospects of spatial biology in colorectal cancer. Front. Oncol. 2024, 14, 1513821. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  43. Elhanani, O.; Ben-Uri, R.; Keren, L. Spatial profiling technologies illuminate the tumor microenvironment. Cancer Cell. 2023, 41, 404–420. [Google Scholar] [CrossRef] [PubMed]
  44. Cilento, M.A.; Sweeney, C.J.; Butler, L.M. Spatial transcriptomics in cancer research and potential clinical impact: A narrative review. J. Cancer Res. Clin. Oncol. 2024, 150, 296. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
Cancers 17 02376 g001
Figure 1. Workflow of AI-HOPE-JAK-STAT for clinical–genomic analysis of the JAK/STAT pathway in colorectal cancer. This figure illustrates the operational framework of AI-HOPE-JAK-STAT, a conversational artificial intelligence system designed to explore JAK/STAT signaling alterations in colorectal cancer (CRC) using natural language input. (a) Users initiate analysis by entering plain language queries such as loading CRC datasets with JAK/STAT pathway genes, filtering by age or mutation status (e.g., JAK3), and requesting survival comparisons between defined cohorts. (b) The system interprets the query through a graphical user interface (GUI) supported by a large language model (LLM), which parses the user’s intent, generates code, and executes the relevant operations. (c) Clinical and genomic data are automatically retrieved and integrated from harmonized datasets. These include patient-level variables and mutation profiles relevant to the JAK/STAT signaling axis. (d) Statistical analyses are then conducted automatically, producing visual outputs such as Kaplan–Meier survival curves and contingency plots. Results are stratified according to user-defined parameters and delivered alongside narrative interpretations, enabling the streamlined investigation of JAK/STAT-driven tumor biology in CRC.
Figure 1. Workflow of AI-HOPE-JAK-STAT for clinical–genomic analysis of the JAK/STAT pathway in colorectal cancer. This figure illustrates the operational framework of AI-HOPE-JAK-STAT, a conversational artificial intelligence system designed to explore JAK/STAT signaling alterations in colorectal cancer (CRC) using natural language input. (a) Users initiate analysis by entering plain language queries such as loading CRC datasets with JAK/STAT pathway genes, filtering by age or mutation status (e.g., JAK3), and requesting survival comparisons between defined cohorts. (b) The system interprets the query through a graphical user interface (GUI) supported by a large language model (LLM), which parses the user’s intent, generates code, and executes the relevant operations. (c) Clinical and genomic data are automatically retrieved and integrated from harmonized datasets. These include patient-level variables and mutation profiles relevant to the JAK/STAT signaling axis. (d) Statistical analyses are then conducted automatically, producing visual outputs such as Kaplan–Meier survival curves and contingency plots. Results are stratified according to user-defined parameters and delivered alongside narrative interpretations, enabling the streamlined investigation of JAK/STAT-driven tumor biology in CRC.
Cancers 17 02376 g001
Cancers 17 02376 g002
Figure 2. AI-HOPE-JAK-STAT analysis of early-onset colorectal cancer patients of ethnicity-specific descent stratified by JAK/STAT pathway alteration status. This figure presents the output of a natural language query executed via AI-HOPE-JAK-STAT to investigate the impact of JAK/STAT pathway alterations on overall survival in early-onset colorectal cancer (EOCRC) patients of Hispanic/Latino descent (H/L). The case cohort includes patients under age 50 with alterations in JAK/STAT pathway genes, while the control cohort includes similarly aged patients of the same ancestry group without such alterations. (a) Pie charts display the proportional representation of selected samples in each cohort relative to the full dataset (n = 5543), highlighting the rarity of JAK/STAT pathway alterations in this subgroup. (b) Kaplan–Meier survival analysis shows a trend toward improved survival in the JAK/STAT-altered cohort (blue) compared to the non-altered group (orange). Although the p-value of 0.0539 from the log-rank test does not reach conventional statistical significance (p < 0.05), the observed difference suggests a possible survival benefit that may warrant further validation in larger cohorts. Shaded confidence intervals illustrate variability in the estimates and highlight the importance of cautious interpretation.
Figure 2. AI-HOPE-JAK-STAT analysis of early-onset colorectal cancer patients of ethnicity-specific descent stratified by JAK/STAT pathway alteration status. This figure presents the output of a natural language query executed via AI-HOPE-JAK-STAT to investigate the impact of JAK/STAT pathway alterations on overall survival in early-onset colorectal cancer (EOCRC) patients of Hispanic/Latino descent (H/L). The case cohort includes patients under age 50 with alterations in JAK/STAT pathway genes, while the control cohort includes similarly aged patients of the same ancestry group without such alterations. (a) Pie charts display the proportional representation of selected samples in each cohort relative to the full dataset (n = 5543), highlighting the rarity of JAK/STAT pathway alterations in this subgroup. (b) Kaplan–Meier survival analysis shows a trend toward improved survival in the JAK/STAT-altered cohort (blue) compared to the non-altered group (orange). Although the p-value of 0.0539 from the log-rank test does not reach conventional statistical significance (p < 0.05), the observed difference suggests a possible survival benefit that may warrant further validation in larger cohorts. Shaded confidence intervals illustrate variability in the estimates and highlight the importance of cautious interpretation.
Cancers 17 02376 g002
Cancers 17 02376 g003
Figure 3. AI-HOPE-JAK-STAT analysis of colorectal cancer patients treated with FOLFOX stratified by JAK/STAT pathway alteration status. This figure presents the results of a natural language-driven analysis conducted through AI-HOPE-JAK-STAT, evaluating overall survival among colorectal cancer (CRC) patients treated with combination chemotherapy (Fluorouracil, Leucovorin, and Oxaliplatin—FOLFOX), stratified by JAK/STAT pathway alteration status. The case cohort includes patients with JAK/STAT alterations who received FOLFOX (n = 282), while the control cohort includes patients without such alterations who received the same regimen (n = 3791). (a) Bar plots summarize the distribution of JAK/STAT pathway alterations across the entire dataset (n = 5543), showing that 10% of patients exhibit alterations in this pathway. The top chart displays absolute counts, while the bottom chart presents percentages. (b) Pie charts illustrate the relative representation of selected patients in each cohort. The case cohort accounts for 5.1% of the dataset, while the control cohort constitutes 68.4%, reflecting the larger proportion of patients without JAK/STAT alterations receiving FOLFOX. (c) Kaplan–Meier survival analysis reveals a statistically significant survival advantage in the JAK/STAT-altered cohort (blue) compared to the control group (orange), with a p-value < 0.0001. The clear separation between curves and narrow confidence intervals suggest a robust association between JAK/STAT pathway alterations and improved treatment outcomes among CRC patients undergoing FOLFOX chemotherapy.
Figure 3. AI-HOPE-JAK-STAT analysis of colorectal cancer patients treated with FOLFOX stratified by JAK/STAT pathway alteration status. This figure presents the results of a natural language-driven analysis conducted through AI-HOPE-JAK-STAT, evaluating overall survival among colorectal cancer (CRC) patients treated with combination chemotherapy (Fluorouracil, Leucovorin, and Oxaliplatin—FOLFOX), stratified by JAK/STAT pathway alteration status. The case cohort includes patients with JAK/STAT alterations who received FOLFOX (n = 282), while the control cohort includes patients without such alterations who received the same regimen (n = 3791). (a) Bar plots summarize the distribution of JAK/STAT pathway alterations across the entire dataset (n = 5543), showing that 10% of patients exhibit alterations in this pathway. The top chart displays absolute counts, while the bottom chart presents percentages. (b) Pie charts illustrate the relative representation of selected patients in each cohort. The case cohort accounts for 5.1% of the dataset, while the control cohort constitutes 68.4%, reflecting the larger proportion of patients without JAK/STAT alterations receiving FOLFOX. (c) Kaplan–Meier survival analysis reveals a statistically significant survival advantage in the JAK/STAT-altered cohort (blue) compared to the control group (orange), with a p-value < 0.0001. The clear separation between curves and narrow confidence intervals suggest a robust association between JAK/STAT pathway alterations and improved treatment outcomes among CRC patients undergoing FOLFOX chemotherapy.
Cancers 17 02376 g003
Cancers 17 02376 g004
Figure 4. AI-HOPE-JAK-STAT analysis of age-stratified colorectal cancer patients with JAK/STAT pathway alterations treated with FOLFOX. This figure summarizes the results of a natural language query executed via AI-HOPE-JAK-STAT to compare survival outcomes and chemotherapy context between colorectal cancer (CRC) patients with JAK/STAT pathway alterations, stratified by age. The case cohort consists of patients younger than 50 years (n = 133), and the control cohort includes patients older than 50 (n = 424). All individuals received combination chemotherapy with Fluorouracil, Leucovorin, and Oxaliplatin (FOLFOX). (a) A histogram displays the age distribution of the full dataset (n = 5543), with a mean age of 57.92 years, confirming the rationale for early- vs. late-onset cohort definitions. (b) Pie charts visualize the proportion of selected samples within each cohort relative to the entire population. The early-onset (case) group comprises 2.4% of samples, while the late-onset (control) group represents 7.6%. (c) Kaplan–Meier survival analysis reveals a statistically significant survival difference between the two age groups, with the early-onset JAK/STAT-altered cohort (blue) demonstrating superior outcomes compared to the older group (orange) (p = 0.0379). (d) A 2 × 2 odds ratio analysis was conducted to assess treatment context by comparing the number of in-context and out-of-context samples (based on receipt of FOLFOX chemotherapy) across cohorts. The stacked bar plot shows that 56.39% of early-onset cases and 48.82% of late-onset controls were in the user-defined treatment context. The resulting odds ratio was 1.356 (95% CI: 0.916–2.006, p = 0.154), suggesting a modest but non-significant enrichment of FOLFOX treatment in the early-onset group.
Figure 4. AI-HOPE-JAK-STAT analysis of age-stratified colorectal cancer patients with JAK/STAT pathway alterations treated with FOLFOX. This figure summarizes the results of a natural language query executed via AI-HOPE-JAK-STAT to compare survival outcomes and chemotherapy context between colorectal cancer (CRC) patients with JAK/STAT pathway alterations, stratified by age. The case cohort consists of patients younger than 50 years (n = 133), and the control cohort includes patients older than 50 (n = 424). All individuals received combination chemotherapy with Fluorouracil, Leucovorin, and Oxaliplatin (FOLFOX). (a) A histogram displays the age distribution of the full dataset (n = 5543), with a mean age of 57.92 years, confirming the rationale for early- vs. late-onset cohort definitions. (b) Pie charts visualize the proportion of selected samples within each cohort relative to the entire population. The early-onset (case) group comprises 2.4% of samples, while the late-onset (control) group represents 7.6%. (c) Kaplan–Meier survival analysis reveals a statistically significant survival difference between the two age groups, with the early-onset JAK/STAT-altered cohort (blue) demonstrating superior outcomes compared to the older group (orange) (p = 0.0379). (d) A 2 × 2 odds ratio analysis was conducted to assess treatment context by comparing the number of in-context and out-of-context samples (based on receipt of FOLFOX chemotherapy) across cohorts. The stacked bar plot shows that 56.39% of early-onset cases and 48.82% of late-onset controls were in the user-defined treatment context. The resulting odds ratio was 1.356 (95% CI: 0.916–2.006, p = 0.154), suggesting a modest but non-significant enrichment of FOLFOX treatment in the early-onset group.
Cancers 17 02376 g004
Cancers 17 02376 g005
Figure 5. AI-HOPE-JAK-STAT analysis of JAK1 mutation status in microsatellite-stable colorectal cancer. This figure illustrates the use of AI-HOPE-JAK-STAT to investigate overall survival differences in colorectal cancer (CRC) patients with stable microsatellite status (MSI-Stable), stratified by JAK1 mutation status. The case cohort includes MSI-Stable patients harboring JAK1 mutations (n = 44), while the control cohort includes MSI-Stable patients without JAK1 mutations (n = 4638). (a) Bar plots display the distribution of MSI types across the full dataset using simplified labels (e.g., X0 = Stable, X1 = Instable). MSI-Stable tumors represent the majority of samples, confirming that the data are sufficient to evaluate JAK1-specific outcomes within this molecular context. (b) Pie charts show the relative proportion of selected samples in the case and control cohorts. The JAK1-mutant group comprises only 0.8% of the dataset, compared to 83.7% for the JAK1 wild-type group within MSI-Stable cases, underscoring the rarity of this alteration. (c) Kaplan–Meier survival analysis reveals no statistically significant difference in overall survival between the two cohorts (p = 0.5691). Although the JAK1-mutant group shows a slightly different survival trajectory, the wide confidence intervals and overlapping curves indicate no clear association with survival in this subset. These results suggest that JAK1 mutation status may not independently influence prognosis among MSI-Stable CRC patients and highlight the importance of considering broader molecular contexts in precision oncology studies.
Figure 5. AI-HOPE-JAK-STAT analysis of JAK1 mutation status in microsatellite-stable colorectal cancer. This figure illustrates the use of AI-HOPE-JAK-STAT to investigate overall survival differences in colorectal cancer (CRC) patients with stable microsatellite status (MSI-Stable), stratified by JAK1 mutation status. The case cohort includes MSI-Stable patients harboring JAK1 mutations (n = 44), while the control cohort includes MSI-Stable patients without JAK1 mutations (n = 4638). (a) Bar plots display the distribution of MSI types across the full dataset using simplified labels (e.g., X0 = Stable, X1 = Instable). MSI-Stable tumors represent the majority of samples, confirming that the data are sufficient to evaluate JAK1-specific outcomes within this molecular context. (b) Pie charts show the relative proportion of selected samples in the case and control cohorts. The JAK1-mutant group comprises only 0.8% of the dataset, compared to 83.7% for the JAK1 wild-type group within MSI-Stable cases, underscoring the rarity of this alteration. (c) Kaplan–Meier survival analysis reveals no statistically significant difference in overall survival between the two cohorts (p = 0.5691). Although the JAK1-mutant group shows a slightly different survival trajectory, the wide confidence intervals and overlapping curves indicate no clear association with survival in this subset. These results suggest that JAK1 mutation status may not independently influence prognosis among MSI-Stable CRC patients and highlight the importance of considering broader molecular contexts in precision oncology studies.
Cancers 17 02376 g005
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Yang, E.-W.; Waldrup, B.; Velazquez-Villarreal, E. Decoding the JAK-STAT Axis in Colorectal Cancer with AI-HOPE-JAK-STAT: A Conversational Artificial Intelligence Approach to Clinical–Genomic Integration. Cancers 2025, 17, 2376. https://doi.org/10.3390/cancers17142376

AMA Style

Yang E-W, Waldrup B, Velazquez-Villarreal E. Decoding the JAK-STAT Axis in Colorectal Cancer with AI-HOPE-JAK-STAT: A Conversational Artificial Intelligence Approach to Clinical–Genomic Integration. Cancers. 2025; 17(14):2376. https://doi.org/10.3390/cancers17142376

Chicago/Turabian Style

Yang, Ei-Wen, Brigette Waldrup, and Enrique Velazquez-Villarreal. 2025. "Decoding the JAK-STAT Axis in Colorectal Cancer with AI-HOPE-JAK-STAT: A Conversational Artificial Intelligence Approach to Clinical–Genomic Integration" Cancers 17, no. 14: 2376. https://doi.org/10.3390/cancers17142376

APA Style

Yang, E.-W., Waldrup, B., & Velazquez-Villarreal, E. (2025). Decoding the JAK-STAT Axis in Colorectal Cancer with AI-HOPE-JAK-STAT: A Conversational Artificial Intelligence Approach to Clinical–Genomic Integration. Cancers, 17(14), 2376. https://doi.org/10.3390/cancers17142376

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop