Table 1.

Applications of artificial intelligence in diagnosis and prediction

Diagnosis and case identification	Prognosis and prediction
Waveform analysis:
Obstetrics – intrapartum monitoring (s31)	Cardiovascular risk prediction (s32-34)
Neurology – remote monitoring of gait (s35)	Prediction of breast cancer survival (s36)
Image analysis:
Pathology – detection of lymph node metastases in breast cancer (s37)	Prediction of outcomes in colorectal cancer (s38)
Dermatology – identification of benign and malignant tumors (s39), identification of fungal infection (s40), classification of skin cancer (s41)	Predicting of survival in non-small cell lung cancer (s42)
Ophthalmology – identification of diabetic retinopathy (s43), grading of macular degeneration (s44)	Prediction of hospitalization due to heart disease (s45)
Cardiology – diagnosis of acute coronary syndrome (s46), identification of heart failure status through remote patient monitoring (s47)	Prediction of primary care utilization (s48)
Radiology – mammography (s49), diagnosis of pneumonia from chest x-ray [10]	Prediction of sepsis in the intensive care unit, emergency department and on the hospital floor (s50)
Electronic health record analysis:	Prediction of central line associated infections and mortality (s51)
Prediction of inpatient diagnosis (s52); Identification of sepsis in the emergency department (s53); Identification of breast cancer symptoms (s54); Heart failure case identification (s55); Identification of patient phenotype from analysis of ICU data (s56); Identification of medical subdomains in clinical notes (s57); Extraction of ICD-10 codes from death certificates and autopsy reports (s58)	Prediction of treatment outcome in social anxiety (s59)
Claims analysis:	Prediction of psychiatric readmission from discharge summaries (s60)
Screening for type 2 diabetes mellitus from payer claims data (s61]

*References s21 to s61 are available in the Online Supplementary Document(Online Supplementary Document).