Table 1

Template use cases for machine learning implementations

RarityScreening for rare conditions where there is a significant clinical and economic benefit from early intervention.Screening an EHR for undiagnosed cardiac amyloidosis,22 familial hypercholesterolaemia,19 or hand-foot-and-mouth disease.23
UrgencyReducing delays in diagnosis or treatment by flagging high-acuity cases or commencing initial management.Reordering the radiologist worklist to prioritise intracranial haemorrhage.24
Automated triage of emergency presentations.25
QuantityDealing with high patient throughput by increasing the speed of clinicians and/or automating routine clinical tasks.Summarising historical notes and identifying relevant clinical data.26
Automated quantification of cardiac volumes on MRI.27
QualityMonitoring care delivery to ensure quality benchmarks are met or flag medical errors.Ensuring patients with high mortality risk receive a palliative care referral at an appropriate point in their admission.28
Double-reading medical imaging to identify missed lesions.29
ComplexityExtending the capabilities of clinicians with advanced diagnostic or treatment decisions on par or exceeding subspecialists.Reinforcement learning for dynamic treatment regimens.30 31
Facial recognition for identification of rare genetic syndromes.32
  • EHR, electronic health record.