TY - JOUR T1 - Artificial Intelligence algorithms cannot recommend a best interests decision but could help by improving prognostication JF - Journal of Medical Ethics JO - J Med Ethics SP - 179 LP - 180 DO - 10.1136/jme-2022-108880 VL - 49 IS - 3 AU - Derick Wade Y1 - 2023/03/01 UR - http://jme.bmj.com/content/49/3/179.abstract N2 - Most jurisdictions require a patient to consent to any medical intervention. Clinicians ask a patient, ‘Given the pain and distress associated with our intervention and the predicted likelihood of this best-case outcome, do you want to accept the treatment?’When a patient is incapable of deciding, clinicians may ask people who know the patient to say what the patient would decide; this is substituted judgement. In contrast, asking the same people to say how the person would make the decision is substituted interests.1 The UK Mental Capacity Act 2005 uses the latter approach, and my comments appertain to that approach.When a patient lacks capacity, the question facing clinicians is, ‘Given the likely outcome of this treatment plan for the patient’s life in the long term, would the patient decide to accept the plan?’ This answer depends on the patient’s wishes, values and other factors they would use when deciding.The proposed development of an algorithm involves the following:2 Studying choices made by healthy people about a variety of interventions, given quantified disbenefits and outcomes.Using information about the people studied, such as age, cultural factors, previous experience, etc, to discover what influence recordable data items have on … ER -