# Healthcare informatics: transparent, explainable and trustworthy AI for health systems

Research synthesis · Healthcare informatics
Canonical page: https://www.narasimharao.net/research/focus-areas/healthcare-informatics/
Author: Narasimha Rao Vajjhala, Professor and Chair of Computer Science, American University in Bulgaria (ORCID 0000-0002-8260-2392)
Updated: 2026-10-04

> Vajjhala and colleagues argue that AI in healthcare succeeds only when it is transparent, accountable and rigorously validated, and they document how machine learning, IoT, 5G and efficiency analysis can support patient care and health-system management.

## Synthesis

Narasimha Rao Vajjhala’s healthcare informatics research, much of it with Philip Eappen, centres on one claim: artificial intelligence improves health systems only when it is transparent, accountable and rigorously validated. [Eappen et al. (2026c)](https://www.narasimharao.net/research/rise-of-ai-in-healthcare-transparent-implementation/) identified professional liability concerns, the lack of explainability in algorithmic decision-making and unclear accountability for AI-influenced clinical outcomes as the key barriers to adoption, and argued that compliance with GDPR and HIPAA must be paired with ethical oversight and collaboration among patients, providers, developers, regulators and insurers. Two edited volumes develop this agenda: one on transparent AI across informatics, patient-centred care and clinical applications ([Eappen et al., 2026a](https://www.narasimharao.net/research/ai-in-healthcare-transparent-approach-informatics/)), and one applying explainable AI to clinical decision-making, telehealth, electronic health records, wearable monitoring and mental health ([Eappen et al., 2026b](https://www.narasimharao.net/research/enhancing-healthcare-informatics-transparent-explainable-ai/)).

The same standard of transparency shapes the group’s empirical work. [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/) trained a lightweight EfficientNetV2B0 network on paediatric chest X-rays under CLAIM and TRIPOD-AI reporting standards and reported an AUC of 0.967 (95% CI 0.953–0.979) with pneumonia recall of 0.98, missing only about 2% of true cases. They reported openly that normal-class recall was lower (0.752) and calibration moderate, framing the false-positive tendency as a clinically preferable trade-off for triage.

Vajjhala and Eappen have also mapped how emerging technologies can support care delivery. They showed that combining machine learning with the Internet of Things supports early disease detection, personalised treatment and operational efficiency, provided that data privacy, model interpretability, bias mitigation and secure connectivity are addressed ([Vajjhala & Eappen, 2024b](https://www.narasimharao.net/research/smart-health-machine-learning-internet-of-things/)), and reviewed how 5G networks can enable smart healthcare applications after COVID-19 exposed overcrowded hospitals ([Vajjhala & Eappen, 2023](https://www.narasimharao.net/research/role-of-5g-networks-healthcare-applications/)). On health-system management, [Vajjhala and Eappen (2024a)](https://www.narasimharao.net/research/data-envelopment-analysis-healthcare-management-review/) showed that data envelopment analysis is widely used for efficiency measurement, benchmarking and resource allocation in hospitals, nursing and outpatient services, while identifying six limitations including input/output selection, sensitivity to outliers and the inability to handle statistical noise.

The crisis dimension runs through the work: an edited volume documented how health informatics, telemedicine and clinical decision support performed during the COVID-19 pandemic and how they could protect patient safety in future crises ([Vajjhala & Eappen, 2022](https://www.narasimharao.net/research/health-informatics-patient-safety-times-of-crisis/)). At the data-science end, [Biba and Vajjhala (2022)](https://www.narasimharao.net/research/statistical-relational-learning-genomics-review/) showed that statistical relational learning suits the relational, noisy and incomplete data of genomics, with the computational cost of inference as its main limitation.

## Key claims with sources

1. Liability concerns, lack of explainability and unclear accountability are the key barriers to healthcare AI adoption ([Eappen et al., 2026c](https://www.narasimharao.net/research/rise-of-ai-in-healthcare-transparent-implementation/)).
2. A lightweight EfficientNetV2B0 model detected paediatric pneumonia with AUC 0.967 and recall 0.98 ([Haveri & Vajjhala, 2026](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/)).
3. Data envelopment analysis supports efficiency measurement and benchmarking in healthcare but is limited by input/output selection, outlier sensitivity and statistical noise ([Vajjhala & Eappen, 2024a](https://www.narasimharao.net/research/data-envelopment-analysis-healthcare-management-review/)).

## The publications and what each contributes

| Publication | Method | Core finding |
|---|---|---|
| [Eappen et al. (2026b)](https://www.narasimharao.net/research/enhancing-healthcare-informatics-transparent-explainable-ai/) | Edited volume of 11 chapters by international contributors; chapters range from ethical and conceptual analysis to systematic review (EEG analysis) and applied explainable-AI system descriptions. | The volume applies explainable and transparent AI across clinical decision-making, telehealth, electronic health records, EEG analysis, wearable monitoring and mental health, pairing ethical analysis with applied systems and strategies for building trust. |
| [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/) | Binary medical image classification with an EfficientNetV2B0 convolutional neural network and ImageNet transfer learning on the Kermany paediatric chest X-ray dataset, following CLAIM and TRIPOD-AI reporting, with bootstrapped confidence intervals, DeLong test, calibration metrics and Grad-CAM. | A lightweight EfficientNetV2B0 model trained under CLAIM and TRIPOD-AI reporting standards achieved an AUC of 0.967 (95% CI 0.953–0.979) and pneumonia recall of 0.98 on paediatric chest X-rays, missing only about 2% of true cases. |
| [Eappen et al. (2026c)](https://www.narasimharao.net/research/rise-of-ai-in-healthcare-transparent-implementation/) | Conceptual and review chapter on transparency and accountability in healthcare AI implementation: barriers to adoption, regulatory compliance (GDPR, HIPAA), multi-actor engagement, and governance recommendations. | The chapter shows that successful healthcare AI depends on transparency, accountability and collaboration among patients, providers, developers, regulators and insurers, and identifies liability, algorithmic opacity and unclear accountability for AI-influenced outcomes as the key barriers to adoption. |
| [Eappen et al. (2026a)](https://www.narasimharao.net/research/ai-in-healthcare-transparent-approach-informatics/) | Edited volume of 11 chapters (4 parts) by international contributors; chapter methods range from conceptual and ethical analysis to case studies and technical deep-learning evaluations. | The volume combines conceptual, ethical, case-based and technical chapters — from transparent implementation and algorithmic scheduling to deep learning for abdominal trauma diagnosis and transparent drug-class prediction — into a framework for transparent healthcare AI. |
| [Vajjhala and Eappen (2024b)](https://www.narasimharao.net/research/smart-health-machine-learning-internet-of-things/) | Review chapter on the convergence of healthcare analytics, machine learning and IoT: benefits, integration challenges and recommended solutions, ethical and security considerations, and emerging advances. | The chapter shows that combining machine learning and IoT supports enhanced patient care, early disease detection, operational efficiency and personalised treatment, and that data privacy, model interpretability, bias mitigation and secure connectivity are prerequisites for deployment. |
| [Vajjhala and Eappen (2024a)](https://www.narasimharao.net/research/data-envelopment-analysis-healthcare-management-review/) | Narrative review chapter consolidating published applications of data envelopment analysis (DEA), a non-parametric efficiency method, in hospital management, nursing and outpatient services. | The review shows that DEA is widely applied in hospital management, nursing and outpatient services for efficiency measurement, benchmarking, resource allocation and performance evaluation, and identifies six limitations, including input/output selection, outlier sensitivity, inability to handle statistical noise and the static nature of traditional models. |
| [Vajjhala and Eappen (2023)](https://www.narasimharao.net/research/role-of-5g-networks-healthcare-applications/) | Review chapter of 5G technologies and their application to smart healthcare services, with a discussion of deployment challenges and leadership and policy recommendations. | The chapter maps 5G technologies to smart healthcare applications, sets out the challenges of deploying them, and gives leadership and policy recommendations for designing 5G-enabled healthcare strategies. |
| [Vajjhala and Eappen (2022)](https://www.narasimharao.net/research/health-informatics-patient-safety-times-of-crisis/) | Edited volume of 13 chapters; chapter methods include machine-learning prediction of coronavirus spread, conceptual design principles for clinical decision support, big data analytics review, registry implementation case study, and national and regional case studies of telemedicine and emergency response. | The volume documents how health informatics performed during the pandemic — machine learning for predicting spread, telemedicine, clinical decision support design, big data analytics, registries and national emergency responses — and shows how improved informatics can protect patient safety in future crises. |
| [Biba and Vajjhala (2022)](https://www.narasimharao.net/research/statistical-relational-learning-genomics-review/) | State-of-the-art literature review (66 references) of probabilistic graphical models and statistical relational learning approaches — probabilistic relational models, relational dependency networks, relational Markov networks — and their genomics applications. | The review shows that statistical relational learning — probabilistic relational models, relational dependency networks and relational Markov networks — suits genomics tasks from sequence annotation to GWAS, and identifies the computational complexity of inference, followed by graph size, as the main limitation shared by most methods. |

## Key terms in this area

- **Multi-actor engagement model:** Collaborative approach to AI implementation involving patients, healthcare providers, AI developers, professional regulatory bodies and health and insurance institutions, which the chapter says promotes transparency by keeping AI systems aligned with human values and clinical needs while maintaining interpretability. (as used in [Eappen et al. (2026c)](https://www.narasimharao.net/research/rise-of-ai-in-healthcare-transparent-implementation/))
- **Barriers to AI adoption:** Obstacles the chapter identifies to adopting AI in healthcare, including concerns about health professional liability, the lack of explainability in algorithmic decision-making, and questions of accountability when AI systems contribute to clinical outcomes. (as used in [Eappen et al. (2026c)](https://www.narasimharao.net/research/rise-of-ai-in-healthcare-transparent-implementation/))
- **Transfer learning:** Adapting a model pretrained on a large dataset (here ImageNet) to a specific task such as chest X-ray classification, useful when annotated clinical data are limited. (as used in [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/))
- **Grad-CAM (Gradient-weighted Class Activation Mapping):** A technique that produces heatmaps highlighting the image regions most influential in a model’s classification decision. (as used in [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/))
- **Expected Calibration Error (ECE):** The weighted average absolute difference between a model’s predicted confidence and its observed accuracy across probability bins. (as used in [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/))
- **Data envelopment analysis (DEA):** A non-parametric method used for evaluating the efficiency of decision-making units. (as used in [Vajjhala and Eappen (2024a)](https://www.narasimharao.net/research/data-envelopment-analysis-healthcare-management-review/))
- **Decision-making unit (DMU):** An entity, such as a hospital or clinic, whose conversion of inputs into outputs is assessed for relative efficiency in DEA. (as used in [Vajjhala and Eappen (2024a)](https://www.narasimharao.net/research/data-envelopment-analysis-healthcare-management-review/))

## Standards, datasets and methods named in these studies

- [General Data Protection Regulation — Regulation (EU) 2016/679](https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng) (Regulation). Regulation (EU) 2016/679 of the European Parliament and of the Council (General Data Protection Regulation). Official Journal of the European Union. Named in [Eappen et al. (2026c)](https://www.narasimharao.net/research/rise-of-ai-in-healthcare-transparent-implementation/).
- [Health Insurance Portability and Accountability Act (HIPAA)](https://www.hhs.gov/hipaa/index.html) (Law). U.S. Department of Health and Human Services. Health Insurance Portability and Accountability Act of 1996 (HIPAA). Named in [Eappen et al. (2026c)](https://www.narasimharao.net/research/rise-of-ai-in-healthcare-transparent-implementation/).
- [Kermany paediatric chest X-ray dataset](https://doi.org/10.1016/j.cell.2018.02.010) (Dataset). Kermany, D. S., Goldbaum, M., Cai, W., Valentim, C. C. S., Liang, H., Baxter, S. L., et al. (2018). Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell, 172(5), 1122–1131.e9. https://doi.org/10.1016/j.cell.2018.02.010 Named in [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/).
- [EfficientNetV2](https://proceedings.mlr.press/v139/tan21a.html) (Model). Tan, M., & Le, Q. V. (2021). EfficientNetV2: Smaller models and faster training. Proceedings of the 38th International Conference on Machine Learning (PMLR 139). Named in [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/).
- [ImageNet](https://doi.org/10.1109/CVPR.2009.5206848) (Dataset). Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009). ImageNet: A large-scale hierarchical image database. 2009 IEEE Conference on Computer Vision and Pattern Recognition, 248–255. https://doi.org/10.1109/CVPR.2009.5206848 Named in [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/).
- [CLAIM — Checklist for Artificial Intelligence in Medical Imaging](https://doi.org/10.1148/ryai.2020200029) (Reporting guideline). Mongan, J., Moy, L., & Kahn, C. E., Jr. (2020). Checklist for Artificial Intelligence in Medical Imaging (CLAIM): A guide for authors and reviewers. Radiology: Artificial Intelligence, 2(2), e200029. https://doi.org/10.1148/ryai.2020200029 Named in [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/).
- [TRIPOD+AI statement](https://pmc.ncbi.nlm.nih.gov/articles/PMC11019967/) (Reporting guideline). Collins, G. S., Moons, K. G. M., Dhiman, P., Riley, R. D., Beam, A. L., Van Calster, B., et al. (2024). TRIPOD+AI statement: Updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ, 385, e078378. https://doi.org/10.1136/bmj-2023-078378 Named in [Haveri and Vajjhala (2026)](https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/).

## References

- Biba, M., & Vajjhala, N. R. (2022). Statistical Relational Learning for Genomics Applications: A State-of-the-Art Review. In S. S. Roy & Y.-H. Taguchi (Eds.), Handbook of Machine Learning Applications for Genomics (pp. 31–42). Springer Nature Singapore Pte Ltd. https://doi.org/10.1007/978-981-16-9158-4_3 — summary and key findings: https://www.narasimharao.net/research/statistical-relational-learning-genomics-review/
- Eappen, P., Gunn, V., Zikos, D., & Vajjhala, N. R. (2026c). The Rise of AI in Healthcare: A Transparent Approach to Implementation. In P. Eappen, V. Gunn, D. Zikos, & N. R. Vajjhala (Eds.), AI in Healthcare: A Transparent Approach to Informatics (pp. 3–16). Chapman and Hall/CRC. https://doi.org/10.1201/9781003624875-2 — summary and key findings: https://www.narasimharao.net/research/rise-of-ai-in-healthcare-transparent-implementation/
- Eappen, P., Gunn, V., Zikos, D., & Vajjhala, N. R. (Eds.). (2026a). AI in Healthcare: A Transparent Approach to Informatics. Chapman and Hall/CRC. https://doi.org/10.1201/9781003624875 — summary and key findings: https://www.narasimharao.net/research/ai-in-healthcare-transparent-approach-informatics/
- Eappen, P., Vajjhala, N. R., Guo, R., Shinners, L., & Gunn, V. (Eds.). (2026b). Enhancing Healthcare Informatics with Transparent and Explainable AI. Auerbach Publications (CRC Press). https://doi.org/10.1201/9781003604440 — summary and key findings: https://www.narasimharao.net/research/enhancing-healthcare-informatics-transparent-explainable-ai/
- Haveri, K., & Vajjhala, N. R. (2026). Deep Learning for Medical Image Analysis: CNN-based Pneumonia Detection on Chest X-Rays. In 2026 6th International Conference on Pervasive Computing and Social Networking (ICPCSN) (pp. 156–161). IEEE. https://doi.org/10.1109/ICPCSN68523.2026.11543623 — summary and key findings: https://www.narasimharao.net/research/cnn-pneumonia-detection-chest-xray-efficientnetv2/
- Vajjhala, N. R., & Eappen, P. (2023). The Role of 5G Networks in Healthcare Applications. In A. Bajpai & A. Balodi (Eds.), Applications of 5G and Beyond in Smart Cities (pp. 87–98). CRC Press. https://doi.org/10.1201/9781003227861-5 — summary and key findings: https://www.narasimharao.net/research/role-of-5g-networks-healthcare-applications/
- Vajjhala, N. R., & Eappen, P. (2024a). Data Envelopment Analysis in Healthcare Management: Overview of the Latest Trends. In Data Envelopment Analysis (DEA) Methods for Maximizing Efficiency (pp. 245–260). IGI Global. https://doi.org/10.4018/979-8-3693-0255-2.ch011 — summary and key findings: https://www.narasimharao.net/research/data-envelopment-analysis-healthcare-management-review/
- Vajjhala, N. R., & Eappen, P. (2024b). Smart Health: Advancements in Machine Learning and the Internet of Things Solutions. In P. Samui, S. S. Roy, W. Zhang, & Y.-H. Taguchi (Eds.), Machine Learning and IoT Applications for Health Informatics (pp. 31–51). CRC Press. https://doi.org/10.1201/9781003424987-3 — summary and key findings: https://www.narasimharao.net/research/smart-health-machine-learning-internet-of-things/
- Vajjhala, N. R., & Eappen, P. (Eds.). (2022). Health Informatics and Patient Safety in Times of Crisis. IGI Global. https://doi.org/10.4018/978-1-6684-5499-2 — summary and key findings: https://www.narasimharao.net/research/health-informatics-patient-safety-times-of-crisis/