There’s a big opportunity for machine learning and big data to bring us cheaper, quality healthcare
Health & Biotech
Health & Biotech
A smart start-up could use machine learning and big data to identify the best — and cheapest — doctors, writes Silicon Quantum Computing director and former Telstra innovation chief Hugh Bradlow.
Over the past 40 years, since the dawn of the computer revolution, industries across the economy have made significant gains in productivity, except for three: government, education and healthcare.
In healthcare, the real cost (in 1989 dollars) of a visit to a GP has risen by 40% between 1989 and 2014, yet the service has not improved in any meaningful way.
In fact, with the increasing corporatisation of primary care medicine, it could be argued that the experience has been degraded due to the limits placed on the time you can spend with the doctor.
On top of that, the Medicare contribution to the cost of the visit during that 25-year period stayed flat (in inflation-corrected dollars) so that by 2014 patients were paying half the cost of the visit out of their own pocket. It is privatisation of the medical system by stealth.
In addition, medicine is not a free market. While doctors can more or less charge what they want without regulatory constraint (some doctors charge up to 10 times the recommended fee for a procedure), from a patient’s perspective there is no readily available data that compares doctors’ fees and their outcomes.
Better healthcae via Internet of Things and Big Data
Finally, as we move from the computer age to the digital age, the Internet of Things (IoT) and Big Data are enabling new diagnostics and treatments to exponentiate.
These have the potential to change the cost structures of medicine dramatically but who is going lead the charge?
There is very little incentive for the medical profession to do so.
To give one example among many, the cost of blood pressure meters has dropped dramatically. These devices are relatively foolproof and do not require trained healthcare professionals to use them.
The US National Institute of Health has just run a trial where they have asked barbers to take the blood pressure of men who come in for a haircut (men are notoriously bad about going to doctors for a check-up).
A nearby pharmacist then visited the barber shop to provide the medications for those who are found to have hypertension.
Within 6 months, the blood pressure of 2/3 of those who followed this procedure had returned to a healthy range, compared to only 11% in a control group who had to visit a physician for treatment.
A new technological opportunity
The upshot is that there is a wealth of new technological opportunity to ‘fix medicine’: make it patient-centric and stop the spiralling costs.
Clearly the biggest opportunity lies in using data to measure not only the performance of the system, but also that of practitioners and the services they provide.
There is already a wealth of data available: billing data, patient records, pathologist data, etc.
Unfortunately, it is fragmented and not properly unified, something that governments have been struggling with for decades.
However, there is a glimmer of hope for a unified electronic health record (EHR).
One of the requirements of Obamacare was for healthcare organisations to collect data in shareable records and the US is now starting to see rapid adoption of EHR usage.
This is a necessary precursor for any significant change in healthcare and clearly regulation, not technology, is the critical enabler of change.
Amazon and Google see opportunities
However, other players, in particular the digital consumer giants, most notably Amazon and Google, are seeing opportunity in this space.
They are organisations that are comfortable building consumer-centric solutions.
Amazon has recently made an interesting move — together with Berkshire Hathaway and JP Morgan they have set up what the Americans call a “HMO” (basically an insurer that contracts their own care providers), initially to service their own staff (between the 3 organisations they have a million personnel so it is a sizeable opportunity for care providers).
The really interesting part, though, is their stated goal to (a) use data to improve healthcare and (b) require providers to make maximum use of digital technology to lower costs.
They have not spelt out their plans, so I can only speculate where this might go, but let me postulate a scenario.
Both Amazon and Google are masters of data and leaders in machine learning.
They could use their accumulated data and create what I shall call a ‘figure of merit’ to determine the effectiveness of doctors and hospitals in performing various procedures.
Automatic comparison of health costs and results
This would enable them to compare costs and outcomes among doctors — and direct patients to those who still perform effectively but charge less.
Doctors are understandably concerned about the imposition of metrics. They have had poor experience based on inadequate metrics imposed on them in the past by politicians looking for quick fixes (such as the four-hour rule for emergency room admissions).
For example, measuring complication rates that don’t consider risk factors is clearly ludicrous — not all patients are equal and the chances of a successful outcome with an obese, elderly smoker are obviously much worse than with a lean 20-year-old athlete.
However, in the 21st century we ought to be able to devise meaningful and valid KPIs.
The whole point about a machine learning algorithm is that, if you have enough data, it will automatically adjust for different circumstances and effectively produce a like-for-like comparison, hence my calling it a ‘figure of merit’.
It will compensate for the patient variability and circumstances under which the operation was performed.
This would be a ‘smart’ KPI that would enable meaningful measurement of the performance of health care professionals, not the clumsy and facile measures that politicians and administrators have tried to impose in the past.
Until someone tries to pioneer such a ‘figure of merit’ we won’t really know how well it work.
But someone must make the first move and my guess is that it won’t be the medical profession but will be some consumer-oriented corporation or a start-up.
Hugh S. Bradlow (@hughbradlow) is President of the Australian Academy of Technology and Engineering. He is also a independent Non-Executive Director of Silicon Quantum Computing.
He was previously Chief Technology Officer and Head of Innovation at Telstra, responsible for the R&D of new technologies and their introduction into Telstra’s business.