The Use of Artificial Intelligence (AI) Technologies in Biomedicine

1. Introduction

Technological advancements are accelerating the integration of artificial intelligence (AI) into biomedicine, but also demonstrate the need for solutions to the complex challenges of healthcare. AI demonstrates an ability to improve precision and efficiency across the spectrum of healthcare: from medical imagining and diagnostics to drug discovery, AI-based technologies are reshaping research and patient care. At the same time, the integration of these technologies into clinical practice requires rigorous validation, transparent algorithms, and rules to ensure ethical implications.

Rapid advancements in medical imaging and artificial intelligence (AI) have opened a new era in healthcare. The integration of these fields has developed applications that support early disease detection, precise diagnosis, personalized treatment strategies, and enhanced patient outcomes [1,2]. Progress in AI technology—particularly deep learning, convolutional neural networks, and generative adversarial networks—has greatly increased the accuracy and efficiency of medical image interpretation [3,4]. These technologies enable the accurate identification of abnormalities, from detecting tumors in radiological scans to recognizing the early indicators of disease in imaging data [5].

In drug discovery and development, AI has the potential to greatly accelerate the development of new drugs compared to traditional methods, which are often time-consuming and have high failure rates [6]. While conventional approaches frequently lack the capacity to fully capture the complexity of biological systems and disease mechanisms, AI-based models can analyze large patient datasets to identify biomarkers, predict treatment responses, and personalize therapies, contributing to the creation of more effective and targeted treatments [7].

Recently, precision medicine has benefited significantly from the integration of multi-dimensional data sources, including genomic information, immunological profiles, and patients’ health records. This progress has been further accelerated by artificial intelligence (AI), which can interpret complex biological datasets and convert them into clinically meaningful insights [8,9]. As a specialized branch of AI, deep learning could detect patterns and anomalies in large datasets, enabling more accurate and predictive healthcare. In genetics, AI algorithms play a key role in rapidly and reliably analyzing genetic markers, providing valuable patterns about patient vulnerabilities and possible treatment responses. Similarly, in immunology, deep learning models can simulate immune system behavior and forecast how it may respond to different therapeutic interventions [10,11,12].

Artificial intelligence (AI) has the potential to fundamentally enhance our understanding and treatment of mental health by integrating multimodal datasets, from genomic profiles and neural circuitry measures to digital behavioral monitoring [13]. This integration can enable the identification of actionable biomarkers and support the development of highly personalized treatment strategies to improve patient outcomes.

2. An Overview of Published Articles

Recent developments in deep learning allow for the assessment of marrow status by using bone marrow biopsy in combination with a visual assessment of [¹⁸F]FDG PET/CT images [14]. For this purpose, a model trained for skeleton segmentation based on the U-Net architecture was retrained for bone marrow segmentation from CT images. The mask obtained from this segmentation, together with the [¹⁸F]FDG PET image, was used to extract radiomics features from which 11 machine learning models for marrow status differentiation were trained. The segmentation model shows excellent values for Jaccard and Dice index of 0.933 and 0.964, respectively. This highlighted the potential of these features for bone marrow assessment, laying the foundation for a new clinical decision support system [14].

Another study explored the application of machine learning to predict the elasticity of RBCs using both image data and detailed physical measurements derived from simulations. The elasticity of red blood cells (RBCs) is crucial for them to fulfill their role in blood, since decreased RBC deformability is associated with various pathological conditions [15]. Simulating RBC behavior in a microfluidic channel generated data on which machine learning techniques were trained. The study highlights the potential of machine learning, including random forests and gradient boosting, to automate and enhance the analysis of RBC elasticity, with implications for clinical diagnostics [15].

Artificial intelligence (AI) has become an important resource for supporting health professionals, particularly by improving tuberculosis (TB) diagnostic workflows. TB is an infectious disease recognized by the World Health Organization as a global emergency and continues to rank among the top ten causes of death worldwide [16]. A study investigated how natural language processing (NLP) techniques and machine learning (ML) models can assist TB diagnosis [16]. Two types of data were examined: free-text information extracted from electronic medical records (EMRs) and structured patient clinical data (CD). Four ML-based strategies were implemented: two models used each data source independently, and two data-fusion models integrated both. The best results were achieved by the model based solely on CD, which reached a sensitivity of 73%, surpassing the typical 40–60% range reported for smear microscopy. These findings highlight the value of analyzing physicians’ narratives and considering their availability alongside structured clinical information [16].

Another study proposed a predictive model for real-time monitoring epidemic trends at the municipal level in Lombardy, northern Italy, by combining Emergency Medical Services (EMS) dispatch data with Geographic Information Systems (GIS) techniques [17]. In contrast with traditional epidemiological models—which depended on official diagnoses and often lacked fine spatial resolution—the approach proposed in [17] used EMS call records as an early indicator of emerging outbreaks. These data were rapidly collected and geo-referenced, and were unaffected by delays in institutional reporting.

The model integrated spatial filtering with machine learning (specifically a random forest classifier) to assign each municipality to one of five epidemic scenarios, ranging from no spread to active transmission with increasing case trends [17]. Developed in partnership with the Lombardy EMS agency (AREU), the system prioritizes operational usability, emphasizing simplicity, speed, and interpretability. Although the phenomenon was complex and the five-class classification added difficulty, the model demonstrated encouraging predictive performance, particularly in identifying areas without active outbreaks [17].

3. Conclusions

Recently, artificial intelligence has demonstrated its potential to revolutionize medical practice, being successfully used from early disease detection and diagnosis to personalized patient treatment. Advanced technologies and innovations in deep neural networks, visual transformers, and large language models will reshape the landscape of healthcare and artificial intelligence.