Ideal algorithms in healthcare: Explainable, dynamic, precise, autonomous, fair, and reproducible

• DOI: 10.1371/journal.pdig.0000006
• Publication Date: 2022-01-18T18:39:51Z
• Authors: Tyler J. Loftus; Patrick J. Tighe; Tezcan Ozrazgat-B...; John P. Davis; Matthew M. Ruppert; Yuanfang Ren; Benjamin Shickel; Rishikesan Kamale...; William R. Hogan; J. Randall Moorman; Gilbert R. Upchurch; Parisa Rashidi; Azra Bihorac
• Keywords: 03 medical and health sciences; 0302 clinical medicine; Computer applications to medicine. Medical informatics; R858-859.7; Review; 3. Good health

Established guidelines describe minimum requirements for reporting algorithms in healthcare; it is equally important to objectify the characteristics of ideal that confer maximum potential benefits patients, clinicians, and investigators. We propose a framework algorithms, including 6 desiderata: explainable (convey relative importance features determining outputs), dynamic (capture temporal changes physiologic signals clinical events), precise (use high-resolution, multimodal data aptly complex architecture), autonomous (learn with minimal supervision execute without human input), fair (evaluate mitigate implicit bias social inequity), reproducible (validated externally prospectively shared academic communities). present an checklist apply highly cited algorithms. Strategies tools such as predictive, descriptive, relevant (PDR) framework, Standard Protocol Items: Recommendations Interventional Trials-Artificial Intelligence (SPIRIT-AI) extension, sparse regression methods, minimizing concept drift can help healthcare achieve these objectives, toward healthcare.