A Meta-Analysis of Artificial Intelligence in the Built Environment: High-Efficacy Silos and Fragmented Ecosystems
Abstract
Highlights
- AI/ML/DL/IoT applications demonstrate substantial performance improvements (15–40%) within specific built environment domains, with a meta-analysis of 71 studies revealing consistent efficacy across energy, water, transportation, construction, and waste management systems.
- Despite technological success, current implementations remain predominantly fragmented, with 91.5% of applications operating as isolated “silos” lacking cross-domain integration (Levels 0 and 1), and only 1.4% achieving real-time integration.
- The proven efficacy of AI-driven solutions within domains provides a strong foundation for scaling smart city implementations, but the lack of integration prevents realization of systemic benefits and synergies.
- Achieving truly connected, sustainable cities demand a paradigm shift from siloed.
- Applications to integrated frameworks that strategically overlay AI-driven intelligence onto existing infrastructure, supported by new governance models and ethical considerations.
Abstract
1. Introduction
1.1. The Smart City Evolution: From Technological Novelty to the Search for Systemic Impact
1.2. The Untapped Potential: AI, ML, DL, and IoT for Overlaying Intelligence on Existing Built Environment Infrastructure
1.3. The Central Research Problem: The Hypothesis of Pervasive Fragmentation
1.4. Aim, Objectives, and Research Questions
- This study is guided by the following key research questions (RQs):
- RQ1: What is the quantifiable efficacy of current AI/ML/DL/IoT applications in improving domain-specific performance and sustainability metrics within various sectors of the built environment?
- RQ2: What is the distribution and historical trend of these smart applications across different built environment domains, particularly concerning their reliance on new versus existing infrastructure?
- RQ3: To what measurable extent do current AI/ML/DL/IoT applications in the built environment demonstrate cross-domain integration rather than operating as fragmented, standalone solutions?
2. Background and Literature Review
2.1. Defining the “True Connected City”: Key Attributes for the Built Environment
- Human-Centricity: The primary goal is enhancing the quality of life for all urban inhabitants and for the professionals who build and maintain the city. Technology is a means to this end, not the end itself.
- Sustainability: Urban systems are designed to minimize their environmental footprint through optimized resource consumption, reduced emissions, and the integration of renewable energy sources.
- Circularity: Embracing circular economy principles, moving from linear “take-make-waste” models to closed-loop systems where resources and materials are reused, recycled, and regenerated.
- Robustness and Resilience: Systems are designed to withstand, adapt to, and recover from shocks and stresses, such as extreme weather events, infrastructure failures, or public health crises.
- Intelligent Infrastructure Leverage: The ability to strategically overlay intelligence upon existing infrastructure assets rather than depending solely on expensive and disruptive new deployments.
- Comprehensive Data Integration: The technical and governance backbone that enables the federation and harmonization of heterogeneous data across different urban domains, forming the foundation for systemic intelligence.
2.2. Review of AI/ML/DL/IoT Application Archetypes in Key Built Environment Domains
2.3. Documented Barriers to System Integration
2.4. Research Gap
3. Materials and Methods
3.1. Study Design and Methodological Framework
3.2. Protocol and Registration
3.3. Data Sources and Search Strategy
3.4. Eligibility Criteria
3.5. Study Selection Process
3.6. Data Extraction and Coding
3.7. Risk of Bias and Study Quality Assessment
- Formula: k = (po − pe)/(1 − pe)
- ○
- po = the relative observed agreement among raters.
- ○
- pe = the hypothetical probability of chance agreement.
- Calculated Values:
- ○
- Title/Abstract Screening: κ = 0.89
- ○
- Full-Text Review: κ = 0.92
- Formula: g = (M1 − M2)/Spooled
- ○
- M1 and M2 are the means of the intervention and control groups, respectively.
- ○
- Spooled is the pooled standard deviation, calculated as: Spooled = ((n1 − 1)s1 + (n2 − 1)s2)/(n1 + n2 − 2)
- n1 and n2 are the sample sizes of the two groups.
- s1 and s2 are the standard deviations of the two groups.
3.8. Heterogeneity: I2 Statistic
- Formula: I2 = 100% × (Q − df)/Q
- ○
- Q is Cochran’s Q statistic, a measure of the total variation.
- ○
- df is the degrees of freedom (number of studies − 1).
- Calculated Value: I2 = 87.3%
3.9. Egger’s Regression Test
- Concept: A linear regression of the standard normal deviate (effect size/standard error) against precision (1/standard error).
- Result: The test was statistically significant (t = 2.14, p = 0.03), indicating the presence of funnel plot asymmetry.
3.10. GRADE Certainty Assessment
4. Results
4.1. Meta-Analysis of Intervention Effectiveness
4.2. Quantitative Analysis of System Integration
4.3. Temporal Trends
4.4. Publication Bias
5. Discussion
5.1. Principal Findings and the “Fragmented Ecosystems”
5.2. Implications for Practice and Policy
- Mandate Open Standards and APIs: City governments must require adherence to open data standards and the provision of public Application Programming Interfaces (APIs) for all new smart infrastructure procurements. This is essential to prevent vendor lock-in and ensure the future interoperability of systems.
- Establish City-Level Data Governance: Municipalities must move beyond siloed departmental data management. Establishing city-level data governance frameworks with clear policies for data sharing, alongside dedicated cross-departmental “integration task forces,” is crucial for breaking down organizational barriers.
- Pilot and Champion Integrated Projects: Public and private investment should be strategically directed towards pilot projects that explicitly target Level 2 or Level 3 integration. Demonstrating tangible benefits in such integrated pilots can de-risk larger-scale implementations and build momentum for broader adoption.
- Develop and Adopt Cross-Domain Metrics: Evaluation frameworks for smart city projects must evolve to capture cross-domain synergies rather than just isolated efficiencies. For example, a new traffic management system should be evaluated not only on traffic flow improvements but also on its impact on city-wide air quality and energy consumption.
- Foster Public–Private Integration Partnerships: Collaborative initiatives are needed to build the “integration infrastructure”—the middleware, platforms, and urban digital twin environments—that can link disparate systems. Such infrastructure should be treated as a public good, supported by public–private partnerships to ensure broad access and utility.
5.3. Theoretical Contributions
5.4. Limitations
5.5. Future Research Directions
- Longitudinal and Scalability Studies: There is a pressing need for long-term research that tracks the performance and societal impacts of smart city interventions over multiple years to determine if initial gains are sustained as projects are scaled from pilot to city-wide deployment.
- Screening and synthesis methods: Utilize AI-assisted screening (active-learning tools) to improve speed, coverage, efficiency and consistency. Conduct a detailed bibliometric content mapping (e.g., VOSviewer 1.6.20) as a complementary lens without diluting the meta-analytic focus.
- Comparative Effectiveness of Integration: Future research should explicitly design studies to directly quantify the added value of integration for instance, by conducting controlled experiments comparing an AI solution implemented in a silo versus the same solution in an integrated context.
- Mixed-Methods Research on Barriers and Enablers: In-depth mixed-methods research combining quantitative data with qualitative case studies is needed to understand how socio-technical hurdles are (or are not) overcome in practice, and what factors most enable successful integration.
- Development of Standardized Benchmarks: The research community would benefit from the development of common benchmarks, open datasets, and shared simulation environments (e.g., open-source urban digital twins [23]) to allow standardized testing and comparison of integrated solutions.
- Ethical, Equity, and Governance Implications: As integration becomes more feasible, research must intensify its focus on the associated ethical and equity dimensions, including issues of data privacy, algorithmic bias, and new models of inclusive governance for large-scale, integrated urban AI systems. We propose also to incorporate government guidance and implementation reports to triangulate practice-led evidence.
6. Conclusions
Author Contributions
Funding
Data Availability Statement
Conflicts of Interest
Abbreviations
Abbreviation | Definition |
AI | Artificial Intelligence |
BIM | Building Information Model |
BMS | Building Management System |
CI | Confidence Interval |
DL | Deep Learning |
RL | Reinforcement Learning |
FL | Federated Learning |
IoT | Internet of Things |
ML | Machine Learning |
PRISMA | Preferred Reporting Items for Systematic Reviews and Meta-Analyses |
PICOS | Population, Intervention, Comparison, Outcome, and Study Design |
SCADA | Supervisory Control and Data Acquisition |
API | Application Programming Interface |
ROBINS-I | Risk Of Bias In Non-randomized Studies of Interventions |
GRADE | Grading of Recommendations, Assessment, Development and Evaluations |
References
- United Nations, Department of Economic and Social Affairs (UN DESA). World Urbanization Prospects: The 2018 Revision—Report; United Nations: New York, NY, USA, 2019; Available online: https://population.un.org/wup/assets/WUP2018-Report.pdf (accessed on 18 July 2025).
- United Nations, Department of Economic and Social Affairs (UN DESA). World Urbanization Prospects: The 2018 Revision—Highlights; United Nations: New York, NY, USA, 2019; Available online: https://population.un.org/wup/assets/WUP2018-Highlights.pdf (accessed on 18 July 2025).
- Gubbi, J.; Buyya, R.; Marusic, S.; Palaniswami, M. Internet of Things (IoT): A vision, architectural elements, and future directions. Future Gener. Comput. Syst. 2013, 29, 1645–1660. [Google Scholar] [CrossRef]
- Albino, V.; Berardi, U.; Dangelico, R.M. Smart Cities: Definitions, Dimensions, Performance, and Initiatives. J. Urban Technol. 2015, 22, 3–21. [Google Scholar] [CrossRef]
- Biyik, C.; Allam, Z.; Pieri, G.; Moroni, D.; O’Fraifer, M.; O’Connell, E.; Olariu, S.; Khalid, M. Smart parking systems: Reviewing the literature, architecture and ways forward. Smart Cities 2021, 4, 623–642. [Google Scholar] [CrossRef]
- Mora, L.; Deakin, M.; Reid, A. Strategic principles for smart city development: A multiple case study analysis of European best practices. Technol. Forecast. Soc. Change 2019, 142, 70–97. [Google Scholar] [CrossRef]
- Zanella, A.; Bui, N.; Castellani, A.; Vangelista, L.; Zorzi, M. Internet of Things for smart cities. IEEE Internet Things J. 2014, 1, 22–32. [Google Scholar] [CrossRef]
- Syed, A.S.; Sierra-Sosa, D.; Kumar, A.; Elmaghraby, A. IoT in smart cities: A survey of technologies, practices and challenges. Smart Cities 2021, 4, 429–475. [Google Scholar] [CrossRef]
- Chen, Y.; Norford, L.K.; Samuelson, H.W.; Malkawi, A. Optimal control of HVAC and window systems for natural ventilation through reinforcement learning. Energy Build. 2018, 169, 195–205. [Google Scholar] [CrossRef]
- Vijayalakshmi, B.; Ramar, K.; Jhanjhi, N.Z.; Verma, S.; Kaliappan, M.; Vijayalakshmi, K.; Vimal, S.K.; Ghosh, U. An attention-based deep learning model for traffic flow prediction using spatiotemporal features towards sustainable smart city. Int. J. Commun. Syst. 2021, 34, e4609. [Google Scholar] [CrossRef]
- Oberascher, M.; Rauch, W.; Sitzenfrei, R. Towards a smart water city: A comprehensive review of applications, data requirements, and communication technologies for integrated management. Sustain. Cities Soc. 2022, 76, 103442. [Google Scholar] [CrossRef]
- Khan, S.; Ali, B.; Alharbi, A.A.K.; Alotaibi, S.; Alkhathami, M.S.; Alshehri, S.; Khattak, H.A.; Khattak, S.; Khan, I. Efficient IoT-Assisted Waste Collection for Urban Smart Cities. Sensors 2024, 24, 3167. [Google Scholar] [CrossRef]
- García López, P.; Montresor, A.; Epema, D.; Datta, A.; Higashino, T.; Iamnitchi, A.; Barcellos, M.; Felber, P.; Rivière, E.; Siganos, G.; et al. Edge-centric Computing: Vision and Challenges. ACM SIGCOMM Comput. Commun. Rev. 2015, 45, 37–42. [Google Scholar] [CrossRef]
- Joyce, A.; Javidroozi, V. Smart city development: Data sharing vs. data protection legislations. Cities 2024, 148, 104859. [Google Scholar] [CrossRef]
- Page, M.J.; McKenzie, J.E.; Bossuyt, P.M.; Boutron, I.; Hoffmann, T.C.; Mulrow, C.D.; Shamseer, L.; Tetzlaff, J.M.; Akl, E.A.; Brennan, S.E.; et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ 2021, 372, n71. [Google Scholar] [CrossRef]
- Sterne, J.A.C.; Hernán, M.A.; Reeves, B.C.; Savović, J.; Berkman, N.D.; Viswanathan, M.; Henry, D.; Altman, D.G.; Ansari, M.T.; Boutron, I.; et al. ROBINS-I: A tool for assessing risk of bias in non-randomised studies of interventions. BMJ 2016, 355, i4919. [Google Scholar] [CrossRef] [PubMed]
- Wolff, R.F.; Moons, K.G.M.; Riley, R.D.; Whiting, P.F.; Westwood, M.; Collins, G.S.; Reitsma, J.B.; Kleijnen, J.; Mallett, S. PROBAST: A tool to assess the risk of bias and applicability of prediction model studies. Ann. Intern. Med. 2019, 170, 51–58. [Google Scholar] [CrossRef]
- Whiting, P.F.; Rutjes, A.W.S.; Westwood, M.E.; Mallett, S.; Deeks, J.J.; Reitsma, J.B.; Leeflang, M.M.G.; Sterne, J.A.C.; Bossuyt, P.M.M. QUADAS-2: A revised tool for the quality assessment of diagnostic accuracy studies. Ann. Intern. Med. 2011, 155, 529–536. [Google Scholar] [CrossRef]
- Egger, M.; Davey Smith, G.; Schneider, M.; Minder, C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997, 315, 629–634. [Google Scholar] [CrossRef]
- Brożek, J.L.; Canelo-Aybar, C.; Compalati, E.; Wiercioch, W.; Baldeh, T.; Khan, R.; Akl, E.A.; Alonso-Coello, P.; Guyatt, G.H.; Schünemann, H.J. GRADE Guidelines 30: The GRADE approach to assessing the certainty of modeled evidence—An overview in the context of health decision-making. BMJ Evid.-Based Med. 2021, 129, 138–150. [Google Scholar] [CrossRef]
- Sipahi, E.B.; Saayi, Z. The world’s first “Smart Nation” vision: The case of Singapore. Smart Cities Reg. Dev. (SCRD) J. 2024, 8, 41–58. [Google Scholar] [CrossRef]
- Baxter, G.; Sommerville, I. Socio-Technical Systems: From Design Methods to Systems Engineering. Interact. Comput. 2011, 23, 4–17. [Google Scholar] [CrossRef]
- Céspedes-Cubides, A.S.; Jradi, M. A review of building digital twins to improve energy efficiency in the operational stage. Energy Inform. 2024, 7, 11. [Google Scholar] [CrossRef]
Criterion | Inclusion Criteria | Exclusion Criteria |
---|---|---|
Population | Systems or processes within the urban built environment: buildings, energy grids, transportation systems, water networks, construction sites, and waste management systems. | Non-urban systems (e.g., agriculture); non-infrastructure applications (e.g., e-governance platforms). |
Intervention | Explicit application of AI, ML, DL, or IoT technology for monitoring, prediction, optimization, or control. | Purely conceptual or theoretical papers; studies not focused on AI/ML/DL/IoT as the primary intervention. |
Comparison | A defined baseline or comparator (e.g., traditional method, pre-intervention state) allowing assessment of the intervention’s effect. | Studies with no comparative data or baseline provided. |
Outcome | Quantitative performance or sustainability metrics reported with sufficient statistical detail for effect size calculation (e.g., mean, standard deviation, n). | Studies reporting only qualitative outcomes or lacking sufficient statistical data for meta-analysis. |
Study Design | Peer-reviewed empirical studies (e.g., journal articles, conference papers) published in English between January 2015 and July 2025. | Literature reviews, editorials, dissertations, theses; non-English language publications; purely simulation studies without real data. |
Source Database | Records Identified | Records After Duplicate Removal | Title/Abstract Screened | Full-Text Articles Assessed | Included in Meta- Analysis (n = 71) |
---|---|---|---|---|---|
MDPI | 1892 | 1446 | 1446 | 112 | 26 |
DOAJ | 1743 | 1169 | 1169 | 95 | 19 |
CORE | 1654 | 1135 | 1135 | 86 | 12 |
BASE | 1287 | 942 | 942 | 62 | 9 |
OpenAIRE | 856 | 584 | 584 | 34 | 5 |
Total | 7432 | 5276 | 5276 | 389 | 71 |
Integration Level | n (% of Studies) | Description |
---|---|---|
Level 0 | 48 (67.6%) | Single-domain, entirely siloed implementation. |
Level 1 | 17 (23.9%) | Limited data sharing (e.g., visualization or one-way data feeds only). |
Level 2 | 5 (7.0%) | Moderate integration (automated data sharing/analysis between 2 and 3 domains). |
Level 3 | 1 (1.4%) | Comprehensive multi-domain integration (real-time co-optimization across domains). |
Key Finding | Core Problem | Strategic Suggestion |
---|---|---|
Domain specific AI solutions yield large improvements (pooled g ≈ 0.92) but operate largely at Levels 0–1 integration (91.5%). | Benefits remain localized, vendor lock-in, siloed governance, and unclear data-sharing rules prevent cross-domain co-optimization. | Mandate open standards and APIs in procurement require implementation-ready data models and interoperability tests at acceptance. |
Algorithmic sophistication has increased (e.g., RL adoption) without tangible gains in integration. | Organizational and economic hurdles outpace technical progress. | Establish city-level data governance and cross-department integration task forces with shared KPIs; publish integration roadmaps. |
Evidence is positive but heterogenous (I2 = 87.3%); possible small-study effects (Egger’s p = 0.03). | Limited longitudinal follow-up and inconsistent reporting obstruct generalization and replication. | Fund Level-2/3 integration pilots with multi-year follow-up, adopt cross-domain metrics |
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. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Alrasbi, O.; Ariaratnam, S.T. A Meta-Analysis of Artificial Intelligence in the Built Environment: High-Efficacy Silos and Fragmented Ecosystems. Smart Cities 2025, 8, 174. https://doi.org/10.3390/smartcities8050174
Alrasbi O, Ariaratnam ST. A Meta-Analysis of Artificial Intelligence in the Built Environment: High-Efficacy Silos and Fragmented Ecosystems. Smart Cities. 2025; 8(5):174. https://doi.org/10.3390/smartcities8050174
Chicago/Turabian StyleAlrasbi, Omar, and Samuel T. Ariaratnam. 2025. "A Meta-Analysis of Artificial Intelligence in the Built Environment: High-Efficacy Silos and Fragmented Ecosystems" Smart Cities 8, no. 5: 174. https://doi.org/10.3390/smartcities8050174
APA StyleAlrasbi, O., & Ariaratnam, S. T. (2025). A Meta-Analysis of Artificial Intelligence in the Built Environment: High-Efficacy Silos and Fragmented Ecosystems. Smart Cities, 8(5), 174. https://doi.org/10.3390/smartcities8050174