Health: reading claims and using a chatbot sensibly
How to read claims about health and risk, what to ask a chatbot about symptoms, and when to see a doctor or call an emergency number at once.
A course for anyone who reads news about medicines and asks a chatbot about symptoms. You will learn to tell strong evidence from weak, make sense of screening results, prepare for an appointment and recognize when you need urgent help. This is not medical advice: the course makes no diagnosis and recommends no treatment, and decisions about your health belong with you and your doctor.
So far there is only this table of contents. Lessons appear one at a time, and we promise no dates. Feed of new lessons (Atom)
15 lessons, 5 parts
Part 1 How to read the evidence
A headline about a new finding rarely says how far to trust it or what it has to do with you.
- coming soon
Levels of evidence
From a single case and a study in cells to a randomized trial: what each level can tell you and what it cannot.
- coming soon
Screening and base rates
What a positive result means when a disease is rare: the ideas of relative risk and base rate from the course on numbers, applied to screening.
- coming soon
When an early finding does not help
What overdiagnosis is, why a screening can find what would never have harmed you, and what to ask your doctor.
Part 2 Medical misinformation
A false claim outlives its correction, especially when someone is earning from it.
- coming soon
The retracted paper and the myth that stayed
Wakefield’s paper on a vaccine and autism was retracted and the fear stayed: why a correction has such trouble catching up with a myth.
- coming soon
Miracle remedies and supplements
The signs of marketing dressed as science, how a supplement differs from a drug in what it must prove, and how to check a claim before you pay.
Part 3 A chatbot and your symptoms
A model knows a lot about medicine, which does not mean a person using it will make a better decision.
- coming soon
What accuracy studies show
Why a model’s high score on textbook cases does not mean the person beside it will decide any better.
- coming soon
A chatbot and urgency
A chatbot can underrate urgency: how that gets tested, what such tests do not show, and why with warning signs it is not the one to ask.
- coming soon
How to ask about symptoms
How to describe the situation fully, ask what must not be missed,, not take a model’s polite agreement for a check.
Part 4 Data about your body
Watches and apps know more and more about you, and reading what they say and protecting it falls to you.
- coming soon
What a watch on your wrist measures
How to tell a signal from a diagnosis, and why an irregular pulse notification is a reason to see a doctor.
- coming soon
Who gets your health data
Medical confidentiality binds your clinic, not most apps and chatbots: what to tell them, and what to check in their terms.
Part 5 The doctor and the emergency line
No course and no chatbot can examine you, so this part is about when and how to go to people who can.
- coming soon
When to see a doctor
What not to keep watching at home: anything new, unexplained, lasting or getting worse, and why a chatbot’s calm answer does not change that.
- coming soon
When to call an emergency number at once
Sudden weakness, trouble speaking, chest pain, severe trouble breathing, thoughts of suicide: whom to call and what to say, with no time lost on a chat.
- coming soon
How to prepare for an appointment
A timeline of symptoms, a list of medicines and your main questions on one page, which a chatbot can help draft without being told more than it needs.
- coming soon
What you may ask your doctor for
Your right to see your records, and how to ask for a plain explanation and a second opinion.
- coming soon
A look ahead: a model with your medical records
Services you hand your medical records to, and the main question: which trials on real patients will show they help rather than harm.