Rajeev S Kapoor via www.linkedin.com
In the past two weeks, we have delved into why physicians are now leaving their professions and taking their children with them. We have seen the negative, sobering statistics of patient-doctor relationships with healthcare reform regulations. We have noticed the move of doctors from medicine to private equity, management, and early retirement.
This week, we look into technology’s role in the physician’s landscape—data collection, diagnosis, even treatment recommendations. Technology is here to stay… and with nearing clinical functionality levels at par with physicians—technology might be here to take over, too.
IBM Watson goes to Medical School
IBM’s Watson supercomputer has been furiously studying all the medical literature that is out there. Combining its cognitive computing capabilities with a capacity to read 65 Million pages of text per second, Watson is now a qualified doctor and is capable of diagnosing diseases based on symptoms and even recommending a treatment.
With the help of natural language processing (NLP), the computer not only pulls out relevant terms to match the search terms in a clinician’s query, but it also understands idioms and some sentences with incorrect English. Experts at IBM are quick to point out that Watson is not just a “search engine on steroids” or even a massive database. In fact, Watson relies on parallel probabilistic algorithms to analyze millions of pages of unstructured text in patient records and the medical literature to locate the most relevant answers to diagnostic and treatment-related questions.
The supercomputer has been furiously “studying” cancer healthcare literature in an effort to find an effective cure for the disease. In line with this pace of growth, more than a dozen leading cancer institutes have collaborated with Watson to identify and personalize treatment options for patients. The institutes will apply Watson’s advanced cognitive capabilities to reduce from weeks to minutes the ability to translate DNA insights, understand a person’s genetic profile and gather relevant information from medical literature to personalize treatment options.
Watson’s training has prepared it initially for a role as a clinical decision support system. But now that it has graduated medical school, it will soon be expected to conclude a diagnostic and treatment regimen with a 95% confidence rating. Imagine the impact this could have on reducing errors in diagnosis !
Digital tools for second opinion
Digital applications are also making breakthrough advances in providing second opinions. Software tools or algorithms for assessing the problem, based on the diagnosis have existed for a while now. Symptom Checkers are interactive programs that allow users to type in the aches, pains and irritations they are experiencing and that may follow up with a series of questions designed to hone in on a disease or condition. Most provide lists of possible diagnoses, usually ranked in order of how likely their algorithm believes they match up to the information provided, rather than a single answer.
Early versions of programs that came out a few years ago did little more than search for key words, but many of today’s symptom checkers are based on sophisticated algorithms that use branching or Bayesian inference — a way of assigning probabilities to hypotheses — that are theoretically supposed to do a better job.
A simple web search can lead to multiple such tools which are available online and can pin point the exact issue. Some of the more reputable ones are run by trusted organizations like Mayo Clinic, American Academy of Pediatrics (designed specifically for children) and WebMD.
However it is worth noting that Harvard Medical School researchers recently conducted an audit of sorts, to test 23 online “symptom checkers”. The study found out that the programs varied widely in accuracy of diagnosis and triage advice, as a whole they were astonishingly inaccurate. Symptom checkers provided the correct diagnosis first in only 34% of cases, and within the first three diagnoses 51% of the time. Needless to say, more work needs to be done to fine tune these algorithms.
Studies show that an estimated 40% of ER visits can be handled over the phone and 50% of ER visits are non-emergencies. A family of 4 could save as much as $2,200 per year with a low deductible plan and up to $4,700 per year with a high deductible plan – all while receiving the same care as they would at a physician’s office in most cases Telemedicine could save all these costs for the patient.
It is clear that the digital disruptors in the healthcare space are changing the way the industry has historically operated. This will only get bigger from here. With the influx of wearable devices, products like AliveCor mobile ECG that is FDA cleared , is available on Amazon, plugs into the smart phone and can detect Arrhythmia with >95% accuracy , thousands of new apps will enable a patient to track each and every vital health stat at their homes from their hand held device.
As many digital leaders have eloquently quoted -“The patient is now becoming the CEO of his own health”
Does this imply the end of the primary care physician profession? Of course not, but this does imply very big changes. Much of what we now see as general practice may well become obsolete, replaced by consumer biometrics and algorithms. What happens to the social aspects of the role, and whether governments will wish to fund this, is far less clear. Specialists will need to rely less on opinion, and more on big data analytics for both diagnosis and delivering treatments. In a disrupting industry like healthcare, only choice is to be disrupted or be the disruptee. So while Physicians are probably getting sick ….of their own jobs and are turning to corporations, management and retirement, they are also confronted with the one of the biggest opportunity of their career- Become a part of the healthcare digital disruption.
Will they let Watson replace them or will they embrace it as their automated assistant to provide help in “medicinal thinking” is yet to be seen.