AI is Transforming Disease Detection

Artificial Intelligence (AI) is revolutionizing healthcare by enabling the development of predictive models based on imaging, electrophysiological, genomic and other data. By analyzing massive amounts of data, AI algorithms can identify patterns that evade any single human’s perception, facilitating personalized prediction and early diagnosis.

Unlike human practitioners, AI can rapidly analyze millions of data, identifying complex associations and patterns that might be impossible for a human to detect. By training on vast datasets, AI systems can discern subtle features associated with various diseases. Moreover, AI’s capacity to handle and interpret large-scale data enables the development of models that can generalize across diverse populations and data modalities. This scalability is crucial for creating robust predictive tools applicable in various clinical settings.

Prominent examples of AI-driven prediction models leverage diverse data types, including electrophysiological, imaging, and genomic data, to enhance disease detection and prognosis. The following examples highlight AI applications utilizing each of these data types.

AI-ECG Risk Estimation for Diabetes Mellitus (AIRE-DM), is a predictive tool, developed by researchers at Imperial College London and Imperial College Healthcare NHS Trust. This AI model analyzes electrocardiogram (ECG) readings to identify individuals at risk of developing type 2 diabetes up to ten years before the condition manifests. By examining subtle changes in routine ECGs, AIRE-DM can predict future risk with approximately 70% accuracy across diverse populations. Early detection enables timely interventions, such as lifestyle modifications, to mitigate the risk of developing type 2 diabetes.

DeepAD is a robust deep-learning model that forecasts disease progression in Alzheimer’s disease. The model integrates high-dimensional MRI features with clinical and demographic information, to predict the future trajectory of patients with Alzheimer’s disease. DeepAD could support early interventions and treatment planning and potentially could impact disease progress.

PrimateAI is another deep-learning model designed to assess the pathogenicity of genetic mutations. Trained on over 300,000 genomic data, this model systematically distinguishes benign from disease-causing mutations. By analyzing deep neural networks, PrimateAI has achieved 88% accuracy in classifying mutations in rare disease patients with developmental, autism spectrum and congenital heart disorders.

AI’s integration into electrophysiological, imaging, genomic and other diagnostic testing, is paving the way for tectonic shifts in how we approach and manage disease. For a large number of conditions, we can expect to be able to better identify those at risk, to detect them earlier, to intervene earlier and, ultimately, to delay onset and improve patient outcomes when disease presents.

References

Chartrand, G., Cheng, P.M., Vorontsov, E., Drozdzal, M., Turcotte, S., Pal, C.J., Kadoury, S., & Tang, A. (2017). Deep learning: A primer for radiologists. Radiographics, 37(7), 2113-2131.

Esteva, A., Chou, K., Yeung, S., Naik, N., Madani, A., Mottaghi, A., Liu, Y., Topol, E.J., & Dean, J. (2019). Deep learning-enabled medical computer vision. npj Digital Medicine, 2(1), 1-9.

Imperial NHS. (2024). AI could predict type 2 diabetes up to 10 years in advance. Imperial College Healthcare NHS Trust. Retrieved from https://www.imperial.nhs.uk/about-us/news/ai-could-predict-type-2-diabetes-up-to-10-years-in-advance

McKinney, S.M., Sieniek, M., Godbole, V., Godwin, J., Antropova, N., Ashrafian, H., Back, T., Chesus, M., Corrado, G.C., & Keane, P.A. (2020). International evaluation of an AI system for breast cancer screening. Nature, 577(7788), 89-94.

Sundaram, L., Gao, H., Padigepati, S.R. et al. Predicting the clinical impact of human mutation with deep neural networks. Nat Genet 50, 1161–1170 (2018). https://doi.org/10.1038/s41588-018-0167-z

Wang, P., Xiao, X., Glissen Brown, J.R., Berzin, T.M., Tu, M., Xiong, F., Hu, X., Liu, P., Song, Y., Zhang, D., & Yang, J. (2022). Development and validation of a deep-learning algorithm for the detection of polyps during colonoscopy. Nature Biomedical Engineering, 6(5), 249-257.

Wen, J., Thibeau-Sutre, E., Diaz-Melo, M., Samper-Gonzalez, J., Routier, A., Bottani, S., Dormont, D., Durrleman, S., & Colliot, O. (2023). Convolutional neural networks for classification of Alzheimer’s disease: Overview and reproducible evaluation. Medical Image Analysis, 44, 101623.