Here’s the plan, in it’s simplest form: deploy medical expert systems using the cell phone network to help people make medical decisions. Use the ability to integrate data from each user into broad statistical patterns to improve the models and plot the progression of epidemics, for example. Then add bicycle-riding computer-aided pharmacists who prescribe basic medications and are paid for using very small health insurance premiums.

Now the detail!

1> Assemble global epidemiological data into a big database. Who gets sick where, what they get sick with, what happens to them. Data will be great in some places, crummy in others, we don’t care right now as long as an estimate of data quality is stored with the data.

2> Assemble simple guides on disease management and health maintenance, translate into 80 languages, including audio versions, and correlate with a symptom database to aid simple diagnosis.

3> From these two resources, assemble a simple (Baysian?) classifier for likely diseases and treatments given symptom input. Where additional questions can distinguish two potential conditions, flag.

4> Prune the tree of all dangerous treatments which have a low chance of success in the field, but keep things which cannot do harm if misapplied.

Result: a map of what’s going on out there, which can be prioritized by effectiveness of intervention and the number of people affected.

One area of this map is the “good intervention” area. This is the place where there are common diseases with effective interventions with low risks if you do something wrong. Typical cases might be non-insulin dependent diabetes, where simple lifestyle changes can have huge health effects. Diarrhea is another area where there’s an effective set of measures which are low risk.

Now crunch this into an expert system which is deployable on cell phones.

The expert system phones home, so the epidemiological database improves rapidly with time – the symptom data which is collected helps refine disease and mortality maps, but many of the illnesses reported cannot be effectively treated without material. The network alone cannot provide antibiotics, and knowing you are dying of a blood infection does not cure it.

However, even this basic system could save millions of lives a year and greatly increase general welfare where available.

The second advancement is the Simple Pharmacopeia. The WHO Essential Medicines List is a likely starting point. Here’s the modification.

1> Filter the drugs in the following ways – no need for refrigeration, no bad effects from over-prescription or likely abuse, less than $10 per course.

2> This probably gives a list of a couple of dozen drugs.

3> Prepare a training course and computer aided (expert system) prescribing guide. Data from this system drives both the logistics and also watches out for things like over / mis-prescription. Comparing all similar villages, one can spot people who’re mis-prescribing, or systemic failures in treatment protocols, lousy training and similar results. Think of the kind of monitoring fast food franchises do of franchise performance, but now make it much, much stronger because all this stuff is safety critical.

4> Villagers pay $10 per year for health insurance. One worker per 1000 villagers, say. Villagers report problems over the network (SMS today, tomorrow interactive diagnostic forms) and the health worker bicycles from village to village with a backpack full of drugs and diagnostic equipment (sugar test strips, pregnancy tests and so on.) They might even collect lab samples.

5> Each village may also have an old lady who gets free insurance for maintain a cell phone and knowing how to operate the expert system, knowing what words like “swelling” mean and how to take a pulse. Her primary job would be to get accurate data into the expert systems in advance of the more highly trained worker arriving, and critically she would be the person who’s job it would be to make people take their full course of antibiotics, by going to their house each morning, handing them their pill, and watching them take it. You can make sure she’s doing her job by, for example, getting her to take a picture of each person taking their pill or something if you’re paranoid, and maybe have a 2D barcode etched on each pill with a laser (in the future.)

6> When the health worker comes, they work through the expert system, doing diagnosis, and then offering whatever treatment they can. Their ability to do more sophisticated diagnosis is largely what drives their job – they can use a stethoscope fairly well, maybe identify where in the abdomen pain is, that kind of thing. All that goes into the machine, which suggests treatment where possible. Where nothing can be done at this level, it becomes a question of referral or palliative care.

7> Things like dentistry, which require special gear, are probably provided by a traveling service. I think there’s a ton of work to be done on low-cost dental technology, however, things like epoxy-based fillings which incorporate a bacteriocidal component to kill whatever is trapped under the filling, and possibly tooth extraction based on bonding things to the tooth to be pulled rather than using pliers.

8> Drugs are sold at no markup to the people who need them. Profit is made on the health cover, not the drug retail, which prevents people in the supply chain looking to make a profit by over-prescribing.

9> $10 per villager for 1000 people == $10,000 budget. $3000 for salary for the worker – this is a skilled job. $3000 for training and equipment and monitoring. $3000 for centralized services (the computer systems,) translations and so on. $1000 profit. 3 billion people on a few dollars a day, multiplied by $1 per person per year profit on health services equals $3bn a year or more in cold, hard cash. There’s room here for a whole new industry, and big business should be looking at this closely.

10> Birth control and sexual health services are a big part of this, but that’s one to discuss another day.

11> Environmental health services (public health, basically) is not a separate field in the developing world. The same health worker who looks in your ear is also the person who trains you how to wash your hands after emptying your composting toilet. It’s all one basic function: protection from illness.

12> The core of this service is accepting that the care is third rate from day one. People will die because we don’t carry the right drugs. They’ll die because the expert system diagnostics are only 80% in practice. They’ll die because the trained worker is sick that day and sends his brother. They’ll die because the software had bugs. We accept that for every ten lives saved, one to three are lost.

It’s this tolerance for a bad healthcare system which allows this system to work at all: you can’t provide 99.9% health care on $10 per year. But you can provide 80% health care, and right now, that’s far, far better access to medical support than the poor can get any other way.

I’m open to talk about this at any time to anybody: hexayurt@gmail.com

PS: a similar expert system plus big databases approach to pandemic flu care is also extremely plausible, and as the model is refined by measured performance (i.e. double blind trials conducted by giving different people different advice and seeing what happens to them) it might actually provide far better care for at least some patients than exposure to the overcrowded and collapsing pandemic-affected hospital environment. The ability to track disease progress through a population is a critical tool in fighting severe pandemics, and providing medical advice by machine helps keep people in their homes, which is critical.