Prabodi Senevirathna

PhD Details

My PhD research focuses on developing data-driven algorithms involving Machine Learning / Artificial Intelligence to quantify and mitigate digital overdiagnosis generated by digital diagnostic algorithms. This involves extracting data from a clinical repository (MIMIC) and applying supervised and unsupervised machine learning methods to answer the research questions. If you are in doubt about what overdiagnosis is, below is a quick overview.

Advances in Machine Learning (ML)/ Artificial Intelligence (AI) tools promise to revolutionize how we diagnose patients. However, this must include a clear understanding of digital health innovations, benefits, and potential adverse effects, mainly when applied to healthier segments of the population. One such negative effect is “Overdiagnosis”. Overdiagnosis is not a false positive; it is a condition that emerges from finding an asymptomatic abnormality that meets existing disease criteria but would have never caused a problem to the individual if left undiagnosed. Overdiagnosis has been extensively described for traditional diagnostic tools and is an emergent theme in digital health. Access to data from healthier individuals via smartphones, smartwatches, and social networks, along with new ML methods, can generate cases of Digital Overdiagnosis by unnecessarily labelling people as sick.