Comprehensive Review on AI-Driven Medical Diagnostics: Current Trends, Techniques, and Future Directions

Abstract

Artificial Intelligence (AI) has significantly transformed the field of medical diagnostics by improving disease detection accuracy, reducing diagnostic time, and enabling continuous patient monitoring. This review paper consolidates findings from 30 research papers published between 2001 and 2025 to map the progress in AI-based diagnostic systems. The paper focuses on five major areas: deep learning for medical imaging, IoT-enabled healthcare systems, Explainable AI (XAI), edge computing, and federated learning. Deep learning architectures such as CNN, ResNet, U-Net, DenseNet, and CheXNet have demonstrated expert-level diagnostic performance in tasks like skin cancer detection, diabetic retinopathy classification, pneumonia detection, and brain tumor segmentation. IoT and wearable technologies further support real-time healthcare monitoring outside clinical environments. Explainable AI improves transparency and clinician trust, while edge AI and federated learning address privacy, latency, and resource limitations in healthcare systems. Despite significant progress, several challenges remain including class imbalance in medical datasets, interoperability issues, high computational requirements, and lack of standardization in explainable frameworks. This paper also identifies future research directions including lightweight AI models, multimodal healthcare systems, and AI deployment in low-resource environments.

Introduction

The text reviews the evolution, applications, benefits, and challenges of Artificial Intelligence (AI)-based medical diagnostic systems, highlighting their growing role in modern healthcare. AI technologies, particularly machine learning and deep learning, have transformed disease diagnosis, medical imaging, patient monitoring, and clinical decision-making by achieving high levels of accuracy and efficiency.

Significant advancements have been demonstrated through landmark studies where deep learning models achieved specialist-level diagnostic performance. Examples include skin cancer classification, diabetic retinopathy detection, and automated pneumonia diagnosis using medical imaging. AI applications have since expanded beyond imaging to include wearable devices, Internet of Things (IoT)-enabled healthcare systems, electronic health records (EHRs), explainable AI (XAI), federated learning, and edge computing.

The proposed AI healthcare framework integrates data from multiple sources, including patient symptoms, medical images (X-rays, CT scans, MRI, retinal scans), and sensor-based wearable devices that monitor physiological parameters such as heart rate, blood pressure, oxygen saturation, temperature, and ECG signals. After data collection, preprocessing and feature extraction are performed to remove noise and improve data quality. Advanced AI models—including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), transformers, and explainable AI techniques—analyze the data to generate disease diagnoses, risk assessments, and personalized healthcare recommendations. The framework supports continuous monitoring, real-time decision-making, and improved patient engagement.

The evolution of AI in medical diagnostics has progressed from early machine learning systems to sophisticated deep learning architectures such as AlexNet, ResNet, Fully Convolutional Networks (FCNs), and U-Net. These models enable automatic feature extraction and have significantly improved disease detection, medical image segmentation, and classification accuracy. Additional developments such as transfer learning, attention mechanisms, and Long Short-Term Memory (LSTM) networks have further enhanced performance across healthcare applications.

Deep learning has become the dominant technology in disease detection, supporting applications such as skin cancer classification, diabetic retinopathy diagnosis, pneumonia detection, brain tumor segmentation, lung cancer prediction, and breast cancer screening. Despite these successes, challenges remain, including imbalanced datasets, high annotation costs requiring specialist expertise, and substantial computational requirements that can limit deployment in resource-constrained environments.

The integration of AI with IoT and wearable technologies has enabled continuous patient monitoring and remote healthcare delivery. Smart devices equipped with sensors can collect real-time physiological data and support early disease detection, particularly for cardiovascular conditions. Benefits include improved accessibility, real-time monitoring, and enhanced healthcare delivery. However, practical limitations include battery constraints, sensor inaccuracies, network dependence, privacy concerns, and deployment costs.

The text emphasizes the importance of Explainable AI (XAI) in healthcare. Clinicians require transparent and interpretable AI systems to trust diagnostic recommendations. XAI techniques help explain model decisions, identify important features, and increase confidence in AI-assisted diagnosis. However, a major challenge remains the trade-off between diagnostic accuracy and interpretability, and there is currently no universally accepted framework for evaluating explainability in healthcare AI systems.

To address privacy and latency concerns, emerging technologies such as federated learning and edge computing have gained importance. Federated learning enables multiple healthcare institutions to collaboratively train AI models without sharing patient data directly, enhancing privacy protection. Edge AI allows diagnostic processing to occur closer to the data source, reducing latency and cloud dependence. Despite these advantages, challenges include hardware limitations, communication overhead, and variability among distributed datasets.

The review also highlights the growing importance of multimodal healthcare systems, which combine medical imaging, electronic health records, wearable sensor data, clinical notes, and IoT-generated information. These integrated systems have the potential to improve diagnostic accuracy and healthcare efficiency but face barriers related to interoperability, fragmented data standards, and infrastructure limitations.

Finally, the paper identifies several critical research gaps, including:

• Data imbalance caused by the rarity of many serious diseases.
• Lack of standardized explainability frameworks for healthcare AI.
• Privacy and security concerns associated with patient data.
• Computational and infrastructure limitations.
• Challenges in integrating heterogeneous healthcare data sources.

Conclusion

AI-powered diagnostics have progressed far beyond the proof-of-concept stage. Over the last decade, deep learning systems have achieved expert-level performance in a vast range of imaging-related tasks. At the same time, the growing use of wearable devices and IoT technologies has enabled continuous health monitoring outside traditional clinical environments. Despite these advances, a clear gap still exists between experimental results and real-world clinical deployment. Interpretability, scalability, integration, and performance in resource-constrained environments are among the key challenges. Overcoming these challenges will require not only technical advances but also improved communication between AI researchers, clinicians, and regulatory bodies. The integration of AI, IoT, edge computing, and federated learning technologies has the potential to transform global healthcare systems and improve medical accessibility worldwide. Future systems must focus not only on accuracy but also on safety, affordability, transparency, and real-world clinical applicability.

References

[1] J. De Fauw et al., \"Clinically applicable deep learning for diagnosis and referral in retinal disease,\" Nature Medicine, vol. 24, no. 9, pp. 1342–1350, Sep. 2018, doi: 10.1038/s41591-018-0107-6. [2] P. Rajpurkar et al., \"Deep learning for specialist-level diagnostic accuracy in medical imaging,\" IEEE Trans. Medical Imaging, vol. 39, no. 6, pp. 1907–1918, Jun. 2020, doi: 10.1109/TMI.2019.2949985. [3] E. J. Topol, \"High-performance medicine: The convergence of human and artificial intelligence,\" Nature Medicine, vol. 25, no. 1, pp. 44–56, Jan. 2019, doi: 10.1038/s41591-018-0300-7. [4] D. Ravi et al., \"Deep learning for health informatics,\" IEEE J. Biomed. Health Informatics, vol. 21, no. 1, pp. 4–21, Jan. 2017, doi: 10.1109/JBHI.2016.2636665. [5] F. Jiang et al., \"Artificial intelligence in healthcare: Past, present and future,\" Stroke Vasc. Neurology, vol. 2, no. 4, pp. 230–243, Dec. 2017, doi: 10.1136/svn-2017-000101. [6] A. Esteva et al., \"Dermatologist-level classification of skin cancer with deep neural networks,\" Nature, vol. 542, no. 7639, pp. 115–118, Feb. 2017, doi: 10.1038/nature21056. [7] V. Gulshan et al., \"Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs,\" JAMA, vol. 316, no. 22, pp. 2402–2410, Dec. 2016, doi: 10.1001/jama.2016.17216. [8] G. Litjens et al., \"A survey on deep learning in medical image analysis,\" Medical Image Analysis, vol. 42, pp. 60–88, Dec. 2017, doi: 10.1016/j.media.2017.07.005. [9] D. Shen, G. Wu, and H.-I. Suk, \"Deep learning in medical image analysis,\" Annual Review of Biomed. Eng., vol. 19, pp. 221–248, Jun. 2017, doi: 10.1146/annurev-bioeng-071516-044442. [10] J. Long, E. Shelhamer, and T. Darrell, \"Fully convolutional networks for semantic segmentation,\" in Proc. IEEE CVPR, Boston, MA, Jun. 2015, pp. 3431–3440, doi: 10.1109/CVPR.2015.7298965. [11] K. He, X. Zhang, S. Ren, and J. Sun, \"Deep residual learning for image recognition,\" in Proc. IEEE CVPR, Las Vegas, NV, Jun. 2016, pp. 770–778, doi: 10.1109/CVPR.2016.90. [12] O. Ronneberger, P. Fischer, and T. Brox, \"U-Net: Convolutional networks for biomedical image segmentation,\" in MICCAI, Munich, Germany, Oct. 2015, pp. 234–241, doi: 10.1007/978-3-319-24574-4_28. [13] P. Rajpurkar et al., \"CheXNet: Radiologist-level pneumonia detection on chest X-rays with deep learning,\" arXiv preprint arXiv:1711.05225, Nov. 2017. [Online]. Available: https://arxiv.org/abs/1711.05225. [14] H. Greenspan, B. van Ginneken, and R. M. Summers, \"Guest editorial deep learning in medical imaging: Overview and future promise,\" IEEE Trans. Medical Imaging, vol. 35, no. 5, pp. 1153–1159, May 2016, doi: 10.1109/TMI.2016.2553401. [15] A. Holzinger et al., \"What do we need to build explainable AI for medical diagnosis?,\" in Proc. CD-MAKE, Regensburg, Germany, Aug. 2017, pp. 189–202, doi: 10.1007/978-3-319-66808-6_19. [16] Z. Obermeyer and E. J. Emanuel, \"Predicting the future — Big data, machine learning, and clinical medicine,\" New England J. Medicine, vol. 375, no. 13, pp. 1216–1219, Sep. 2016, doi: 10.1056/NEJMp1606181. [17] A. Krizhevsky, I. Sutskever, and G. E. Hinton, \"ImageNet classification with deep convolutional neural networks,\" in Advances in NeurIPS, Lake Tahoe, NV, Dec. 2012, pp. 1097–1105. [18] I. Kononenko, \"Machine learning for medical diagnosis: History, state of the art and perspective,\" Artificial Intelligence in Medicine, vol. 23, no. 1, pp. 89–109, Aug. 2001. [19] N. Tajbakhsh et al., \"Convolutional neural networks for medical image analysis: Full training or fine tuning?,\" IEEE Trans. Medical Imaging, vol. 35, no. 5, pp. 1299–1312, May 2016, doi: 10.1109/TMI.2016.2535302. [20] X. Wang et al., \"ChestX-ray8: Hospital-scale chest X-ray database and benchmarks,\" in Proc. IEEE CVPR, Honolulu, HI, Jul. 2017, pp. 3462–3471, doi: 10.1109/CVPR.2017.369. [21] G. Huang, Z. Liu, L. van der Maaten, and K. Q. Weinberger, \"Densely connected convolutional networks,\" in Proc. IEEE CVPR, Honolulu, HI, Jul. 2017, pp. 2261–2269, doi: 10.1109/CVPR.2017.243. [22] Z. Zhou et al., \"UNet++: Redesigning skip connections to exploit multiscale features in image segmentation,\" IEEE Trans. Medical Imaging, vol. 39, no. 6, pp. 1856–1867, Jun. 2020, doi: 10.1109/TMI.2019.2959609. [23] B. Shickel, P. J. Tighe, A. Bihorac, and P. Rashidi, \"Deep EHR: A survey of recent advances in deep learning techniques for electronic health record (EHR) analysis,\" IEEE J. Biomed. Health Informatics, vol. 22, no. 5, pp. 1589–1604, Sep. 2018, doi: 10.1109/JBHI.2017.2767063. [24] A. Y. Hannun et al., \"Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network,\" Nature Medicine, vol. 25, no. 1, pp. 65–69, Jan. 2019, doi: 10.1038/s41591-018-0268-3. [25] S. M. McKinney et al., \"International evaluation of an AI system for breast cancer screening,\" Nature, vol. 577, no. 7788, pp. 89–94, Jan. 2020, doi: 10.1038/s41586-019-1799-6. [26] D. Ardila et al., \"End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography,\" Nature Medicine, vol. 25, no. 6, pp. 954–961, Jun. 2019, doi: 10.1038/s41591-019-0447-x. [27] G. Luo et al., \"A comprehensive study on hybrid U-Net architectures with attention gates for automated medical image segmentation,\" IEEE Access, vol. 9, pp. 18595–18607, 2021, doi: 10.1109/ACCESS.2021.3064755. [28] M. Havaei et al., \"Brain tumor segmentation with deep neural networks,\" Medical Image Analysis, vol. 35, pp. 18–31, Jan. 2017, doi: 10.1016/j.media.2016.05.004. [29] S. Hochreiter and J. Schmidhuber, \"Long short-term memory,\" Neural Computation, vol. 9, no. 8, pp. 1735–1780, Nov. 1997, doi: 10.1162/neco.1997.9.8.1735. [30] S. S. M. Salehi, D. Erdogmus, and A. Gholipour, \"Tversky loss function for image segmentation using 3D fully convolutional deep networks,\" in Proc. MLMI, 2017, pp. 379–387, doi: 10.1007/978-3-319-67389-9_44. [31] H. R. Roth et al., \"DeepOrgan: Multi-level deep convolutional networks for automated pancreas segmentation,\" in MICCAI, 2015, pp. 556–564, doi: 10.1007/978-3-319-24553-9_68. [32] D. Bahdanau, K. Cho, and Y. Bengio, \"Neural machine translation by jointly learning to align and translate,\" in Proc. ICLR, 2015. [33] J. Schmidhuber, \"Deep learning in neural networks: An overview,\" Neural Networks, vol. 61, pp. 85–117, Jan. 2015, doi: 10.1016/j.neunet.2014.09.003.

Copyright

Copyright © 2026 Shivam Devkar, Prof. Sachin Takale, Jayesh Bhosale, Shreya Malhan, Tanika Khandelwal. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.