IB Digital Society 4.4A Medicine and Health Notes

4.4A Medicine and Health

1. The organising concept is systems: care is a loop (symptoms, data, diagnosis, treatment, monitoring) and a change at one point ripples through the rest.
2. Keep impacts (changes already here, e.g. faster diagnosis) apart from implications (wider consequences, e.g. privacy risks, unequal access).
3. Content topics that meet here: data and AI (diagnosis, prediction), networks (care across distance), robots (surgery); values and ethics run throughout because health is high-stakes.

Diagnosis by Algorithm

1. Clinical decision support systems use AI and machine learning to read medical data, spot patterns, and suggest what a clinician should look at next.
2. Now mainstream: the US FDA has authorised well over a thousand AI-enabled devices, around three-quarters in medical imaging (X-rays, CT, retinal scans).
   1. Some tools match or beat specialists at narrow tasks such as flagging a tumour or diabetic eye damage.
3. The design principle is human-in-the-loop: the AI narrows attention to what matters, then a clinician makes the call. This creates a power question about how far a doctor should trust an opaque system. It is also a systems risk, because a confident wrong answer can propagate if nobody checks it.
4. Impact: earlier, faster, sometimes more accurate diagnosis. Implication: automation bias (deferring to the machine) and unclear accountability when the tool is wrong.

The Monitored Body

1. Wearable health devices and remote monitoring turn the body into a continuous stream of data, so care no longer waits for a clinic visit.
2. Examples: smartwatches detect irregular heart rhythm; over-the-counter continuous glucose monitors; remote patient monitoring sends home readings to a care team between appointments.
3. Through space, the home becomes a clinical space; through data, the constant stream is valuable but intimate and exposed.
4. Impact: earlier warnings, more engaged patients. Implication: anxiety from self-tracking, false alarms, and who else sees the data (insurers, device makers).

Reaching Patients: Telemedicine and Access

1. Telemedicine delivers care over a network (video, messaging, connected devices); it moved mainstream after 2020, with AI symptom-checkers and chatbots as a first point of contact.
2. Through space it shrinks distance. But it depends on a device, connectivity, and digital confidence, so patients who lack those can be shut out, widening the digital divide. Competing people: rural and elderly patients, health systems, insurers.
3. Impact: care reaches more people at lower cost per visit. Implication: who gets left out, how far a diagnosis without a physical exam can be trusted, and how reliable direct-to-patient AI advice is.

When Health Tech Gets It Wrong

1. A medical algorithm learns from past data, so algorithmic bias in that data is carried straight into care.
2. A widely used US care-referral algorithm was found in 2019 to under-refer Black patients because it used past spending as a proxy for need.
   1. Pulse oximeters read less accurately on darker skin. Clinical LLM chatbots can state a confident falsehood, which turns dangerous at a bedside.
3. Through values and ethics: fairness, consent, and accountability when software harms a patient. Through power: an illness data trail is private yet valuable to insurers, employers, advertisers.
4. Because harms are documented and trust erodes easily, testing across all groups, transparency, and human oversight matter.

Weighing a Contested Balance

1. Digital medicine has real benefits and real costs. The benefits are more accurate, faster, more available care. The costs include encoded bias, exposed private data, and over-reliance.
2. Contested: whether a tool heals or harms depends on how people build and govern it (testing across groups, clinician in the loop, data protection, independent validation). Viewed as a system, its outcome is set by those design and governance choices, not by the technology alone.