How will AI help my doctor?

Episode Summary

Executive Summary: This Tech Life episode examines how AI is reshaping healthcare, from improving diagnosis and freeing clinicians from admin to enabling prevention and better care in low-resource settings. It also highlights health-tech equity: a student-built mobility vest for blind users, a breakthrough in tracing a child abuse victim, and a new femtech app designed to be free, private, and accessible.

Main Topics: AI in healthcare: promise and limits (Priority: 5/5): Dr William Waugh explains why medicine is a difficult test for AI because of regulation, life-or-death stakes, and the human doctor-patient relationship, while noting AI can improve diagnosis and efficiency. From reactive care to prevention (Priority: 5/5): The discussion argues that healthcare should shift from treating late-stage illness to detecting early warning signs through blood tests, wearables, and other data sources to prevent or delay disease. Equity and access in health tech (Priority: 4/5): The episode warns that new health technologies may mainly benefit wealthy users unless governments and developers ensure access for higher-risk and lower-income populations. AI and healthcare in low- and middle-income countries (Priority: 4/5): AI could help community health workers in places with few doctors by giving them decision support based on the latest medical research and case data. Student-led assistive technology for visual impairment (Priority: 3/5): A West Bank student startup, Guidexa, is building an affordable smart safety vest with sensors, GPS, and motion tracking to help blind and visually impaired people navigate safely. Privacy-first femtech and menstrual health (Priority: 5/5): Amber Vaudigel argues that period-tracking apps often monetize women’s data or lock features behind paywalls, and presents 28X as a free, private, clinically informed alternative. Good news from cyber safety investigations (Priority: 4/5): The BBC reports that the Internet Watch Foundation helped identify and trace a long-unidentified victim of child sexual abuse imagery after an analyst recognized a school uniform and badge.

Key Arguments: AI is hardest to deploy in healthcare because it must meet strict regulation, handle life-or-death decisions, and preserve the human relationship between clinicians and patients. AI can still outperform humans in narrow tasks such as analyzing large volumes of scan images and can help doctors make more accurate diagnoses when supervised by a human. Automating note-taking and record-keeping could reduce administrative burden and let doctors spend more time with patients instead of screens. Healthcare systems should move from a reactive model focused on hospitals and late-stage sickness toward prevention and early detection. Wearables and biological data may reveal disease risk years before symptoms appear, creating opportunities for earlier intervention. Without regulation, health tech may widen inequality by serving affluent, health-conscious users rather than those at highest risk. In low-resource countries, AI could amplify the capabilities of community health workers and partially offset shortages of doctors and nurses. Femtech apps often force women into a bad trade-off between paying subscriptions and surrendering intimate data to advertisers. A privacy-first architecture that keeps data on-device can reduce surveillance risk, hacking risk, and pressure from governments or advertisers. Assistive tech can be affordable and student-led while still delivering meaningful independence for disabled users. Investigative cyber work can produce real-world safeguarding outcomes when analysts connect online evidence to a victim’s identity.

Data Points: Smartwatch users worldwide: about half a billion - Dr Waugh cites the scale of wearable adoption as evidence that people are already tracking health data. Health budget spent on hospitals: 90% - Dr Waugh says most modern health spending goes to late-stage sickness rather than prevention. Disease incubation time: 15 to 20 years - He notes heart disease and neurodegenerative conditions can develop long before symptoms appear. Guidexa obstacle detection range: several meters ahead - The student-built vest uses distance sensors to warn visually impaired users of obstacles. IWF first saw the victim in abuse material: 2020 - Joe Tidy explains the girl’s images and videos had been appearing for years. Time the victim had been appearing in material: about five and a half years - The IWF had repeatedly encountered the same victim across investigations. Estimated age in the abuse material: 13 or 14 - Investigators believe she was a young teenager when the images were made. Time from identification to police confirmation: within three days - After the analyst reported the likely identity, police traced the victim quickly. 28X rollout countries: 100 countries - Amber Vaudigel says the app will be internationally available and expanded widely. 28X languages: 35 languages - The app aims to reach users across many language groups. 2 million pregnant women a day: 2 million - Vaudigel describes the scale of users on her previous Pregnancy Plus platform.

Pivotal Quotes: "Healthcare may be the hardest test because you have layers of regulation, you have real life and death scenarios." — Dr William Waugh: Explaining why medicine is uniquely challenging for AI adoption. "We are building with 28x the first free, completely privacy-safe, but clinically-informed period tracker that keeps all your personal health data exactly where it belongs and is on your phone, so not on servers." — Amber Vaudigel: Describing the design philosophy behind her new femtech app. "It was a really lovely outcome to something that we've been seeing for years and one that we don't always get a result." — Mabel (IWF analyst): Reflecting on the successful identification and safeguarding of the abuse victim.

Implications: The episode suggests AI can improve care, but only if privacy, equity, and human oversight are built in. Health tech’s future will depend on whether it serves everyone—not just wealthy users—and whether it protects sensitive data by design.