Locked And Adaptive Algorithms In Healthcare: Differences, Importance And Regulatory Hurdles

Consider two hypothetical artificial intelligence (A.I.)-based assistive tools in a healthcare setting: algorithm A and algorithm B. Algorithm A, having been rigorously trained on existing data sets, will provide robust results on what is known in those data inputs.  On the other hand, algorithm B, in addition to having been rigorously trained, analyses data in real-time from around the world to get new insights and improve on its recommendations. Would you pick one over the other? Arguably, the second type of A.I., relying on an “adaptive” algorithm, holds more potential; especially in cases of new or poorly understood conditions. Take for example the COVID-19 pandemic, such an adaptive A. I. tool could pool data from institutions using the same system and have better insights into how specific dosage influences a patient’s health.  So far, type A algorithms have been put into practice and approved by regulatory bodies like the FDA. But health authorities are already anticipating a future with adaptive algorithms. In 2019, the FDA, which has been taking the lead in regulating A. I.-based medical tools, issued a whitepaper to propose a regulatory framework for these algorithms and to receive feedback from stakeholders. They followed up with an action plan in 2021, laying out the pathway for approvals. However, this only represents the early steps in regulating adaptive medical A.I. These will most certainly form part of the future of medical settin...
Source: The Medical Futurist - Category: Information Technology Authors: Tags: TMF Artificial Intelligence in Medicine A.I. in healthcare locked algorithm adaptive algorithm Source Type: blogs