ideasIdea
AI-first reading for high-volume common cancer diagnoses, pathologist for the exceptions
Let validated AI make the first read on routine, high-volume samples like cervical smears and standard breast biopsy stains, so scarce pathologists spend their time on the difficult cases.
Pathology AI is now good enough for narrow, high-volume tasks: cervical cytology screening, prostate biopsy detection, HER2 and ER scoring on breast biopsies. In systems with a fraction of the needed pathologists, an AI-first workflow with pathologist sign-off only on flagged or discordant cases could multiply capacity. The unsolved problems are local validation on different scanners and populations, regulatory acceptance in each jurisdiction, and liability.
Hypothesis
In a pathologist-scarce setting, an AI-first workflow will at least triple cases reported per pathologist-hour while keeping sensitivity for malignancy above 98% on prospective audit.
Rationale
AI cervical screening tools have already shown non-inferiority to cytotechnologists in several settings; the same triage logic is used for tuberculosis chest X-rays in high-burden countries.
What would test it
A prospective, paired-read study in two LMIC laboratories on cervical cytology and breast core biopsies: AI-first with sign-off versus full human read, measuring throughput, sensitivity, specificity, and time to report.
Maturity
early clinical
Who has to act
data
Cost to try
Medium ($1M to $50M)
Years to first evidence
4
Bottlenecks it attacks
- Most of the world has almost no cancer care · Seven in ten cancer deaths happen in low- and middle-income countries, where radiotherapy, pathology, surgery and drugs are scarce.
- Not enough oncologists, nurses, pathologists, physicists · The number of people with cancer is rising faster than the workforce trained to treat them.
- AI that is built but not validated or deployed · Thousands of cancer AI models are published; a handful are in clinical use, and fewer have shown they help patients.