OBJECTIVES: To explore the extent of actual use of artificial intelligence (AI)/ machine leraning (ML) protocols for diagnosing cancer in prospective settings, given that numerous AI/ ML protocols have shown promising results in cancer diagnosis in validation tests involving retrospective patient databases

METHODS: We searched PubMed for studies that used AI/ML protocols for cancer diagnosis in prospective (clinical trial/ real-world) setting, from inception till 17th May 2021. We looked for studies in which the AI/ML diagnosis actually aided clinical decision making. Data pertaining to the cancer, patients, and the AI/ML protocol were extracted. Comparison of AI/ML protocol diagnosis with human diagnosis was recorded. Through a post-hoc analysis, data from studies describing validation of various AI/ML protocols was extracted.

RESULTS: Only 18/960 initial hits (1.88%) utilised AI/ML protocols for diagnostic decision-making. Most protocols used ANN (artificial neural network) and DL (deep learning). AI/ML protocols were utilised for cancer screening, pre-operative diagnosis, pre-operative staging, and intra-operative diagnosis of surgical specimen. The reference standard for 17/18 studies was histology. AI/ML protocols were used to diagnose cancers of the colorectum, skin, uterine cervix, oral cavity, ovaries, prostate, lungs, and brain. AI/ML protocols were found to improve human diagnosis, and had either similar or better performance than the human diagnosis, especially that made by the less experienced clinician. Validation of AI/ML protocols was described by 223 studies. There was a huge variation in the number of items used for validation.

CONCLUSIONS: A meaningful translation from validation of AI/ML protocols to their actual usage in cancer diagnosis is lacking. Development of regulatory framework specific for AI/ML usage in healthcare is essential.