Electronic Health Records

News

Is crowdsourcing the new frontier of clinical decision support?

By Jennifer Bresnick

- It may seem like the province of comic book artists and gadgeteers hoping to raise some money for new projects through portals like Kickstarter and Indiegogo, but crowdsourcing might be coming to a hospital near you.   Harnessing the power of the people is no new concept, but allowing amateur medical sleuths to use their insight and Google power to suggest answers to hard-to-diagnose cases may be a first for medicine, which traditionally relies on first-hand examinations and highly trained professionals to do most of the heavy lifting.  The idea is taking off, however, with participation from startups like CrowdMed and the backing of some physicians and traditional researchers who think the method can reduce hospitalizations, slash costs, and save lives.

The CrowdMed slogan “you don’t need a medical degree to help save a life” might rub some physicians the wrong way, but crowdsourcing can be seen as a mash up of clinical decision support and big data analytics, both of which are providing innovative solutions for patients at an unprecedented rate.  Physicians are increasingly relying on computerized applications that generate lists of possible diagnoses based on clinical input, adjusting their predictions based on new symptoms or test results and mining data produced by the patient’s EHR.  Why not add human intelligence to the process, too?

“There is understandably some apprehension about letting the lay public in on medical research or even assisting with making medical diagnoses because the stakes are so high in medicine,” admits Benjamin Ranard, a medical student at the University of Pennsylvania and author of a study that used a public contest to record the locations of more than 1,400 defibrillators, raising awareness of the life-saving equipment.  “However, studies we reviewed showed that the crowd can be very successful, such as solving novel complex protein structure problems or identifying malaria infected red blood cells with a similar accuracy as a medical professional.”

“Groups hold far more knowledge collectively than any individual member, no matter how brilliant,” asserts the CrowdMed website.  “No single individual, even a doctor, can keep track of thousands of unique medical disorders. Further, physicians are accustomed to recognizing and diagnosing common diseases, not rare ones.”  The more people searching for an answer, the more likely the answer will be found in a timely manner, the website insists, and the cheaper the bill for the patient.  CrowdMed boasts of having saved patients more than $3.8 million so far due to shorter hospitalizations, fewer specialist consults, and fewer unnecessary tests and prescriptions.

But for every zebra following the hoof beats, there are plenty more horses, argues Dr. Joshua Liao of Brigham and Women’s Hospital.  “Chest pain that for all the world sounds like a heart attack sometimes turns out to be bad heartburn.  And not everything seen on an X-ray or other imaging is clinically important,” he explains.

“Diagnosis is shaped by sensory observation, and crowdsourcing methods miss this vital information. All doctors are taught – on day one of medical school, in most cases – that the most important thing in diagnosis is a good history and physical examination.  Because groups such as CrowdMed must rely on what patients report when they input their information, their online ‘MDs’ are left to diagnose with potentially incomplete or incorrect information.”

“What can clarify can also confuse. What is meant to reduce cost can also create them: the expense of added tests and treatments necessitated by crowdsourced diagnoses that turn out to be wrong,” Liao says.  Sometimes too many cooks can spoil the broth, and when lives are at stake, not every physician is willing to trust the “hive mind” of the Internet.

If crowdsourcing is to become an effective tool in the fight against rare diseases and skyrocketing medical costs, it will need to be harnessed and refined the way machine learning clinical decision tools are currently being perfected, and provide a reliable, useful supplement to the physician’s finely honed skills.

Continue to site...