Robust diagnostic classification via Q-learning
| Autor: | Krishna Somandepalli, Shuting Zheng, Victor Ardulov, Emma Salzman, Victor R. Martinez, Shrikanth S. Narayanan, Somer L. Bishop, Catherine Lord |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Computer science
Process (engineering) Autism Spectrum Disorder Science Autism Intellectual and Developmental Disabilities (IDD) Clinical Decision-Making Q-learning Decision Support Systems Context (language use) Machine learning computer.software_genre Article Task (project management) Machine Learning 03 medical and health sciences Clinical 0302 clinical medicine Theoretical Robustness (computer science) Models Clinical Research Diagnosis Reinforcement learning Humans 0501 psychology and cognitive sciences Medical diagnosis Interpretability Pediatric Multidisciplinary business.industry 05 social sciences Health care Models Theoretical Decision Support Systems Clinical Brain Disorders Mental Health Attention Deficit Disorder with Hyperactivity Medicine Artificial intelligence business computer 030217 neurology & neurosurgery Algorithms Software 050104 developmental & child psychology |
| Zdroj: | Scientific reports, vol 11, iss 1 Scientific Reports Scientific Reports, Vol 11, Iss 1, Pp 1-9 (2021) |
| Popis: | Machine learning (ML) models have demonstrated the power of utilizing clinical instruments to provide tools for domain experts in gaining additional insights toward complex clinical diagnoses. In this context these tools desire two additional properties: interpretability, being able to audit and understand the decision function, and robustness, being able to assign the correct label in spite of missing or noisy inputs. This work formulates diagnostic classification as a decision-making process and utilizes Q-learning to build classifiers that meet the aforementioned desired criteria. As an exemplary task, we simulate the process of differentiating Autism Spectrum Disorder from Attention Deficit-Hyperactivity Disorder in verbal school aged children. This application highlights how reinforcement learning frameworks can be utilized to train more robust classifiers by jointly learning to maximize diagnostic accuracy while minimizing the amount of information required. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|