In order to determine treatment for a patient's Parkinson's disease, doctors need to know the severity of their symptoms. Unfortunately, gauging that severity isn't an exact science. A new machine learning-based app, however, is intended to make it one.

Ordinarily, Parkinson's symptoms such as tremors and walking difficulties are assessed when patients visit a specialist at a clinic – something that happens a few times a year. Not only can the assessment vary with the specialist performing it, but it's also possible that the symptoms may be unusually mild or severe on the days of the visits.

Doctors will also sometimes have patients keep a diary of their symptoms, but again, patients may not be accurate or consistent when rating the severity of those symptoms.

With that in mind, scientists from Johns Hopkins University, the University of Rochester Medical Center and England's Aston University created an app that's designed to do the job objectively. Called HopkinsPD, it utilizes the smartphone's microphone, touchscreen and accelerometer to assess patients' symptoms as they perform tasks that involve voice sensing, finger tapping, gait measurement, balance and reaction time.

Patients can use the app to test themselves when and wherever they want, receiving a Parkinson's disease severity score when they're done. That score can then be shared with their doctor.

"If you think about it, it sounds crazy, but until these types of studies, we had very limited data on how these people function on Saturdays and Sundays because patients don't come to the clinic on Saturdays or Sundays," says Rochester neurologist E. Ray Dorsey, who led the research along with Johns Hopkins assistant professor of computer science, Suchi Saria. "We also had very limited data about how people with Parkinson's do at two o'clock in the morning or 11 o'clock at night because unless they're hospitalized, they're generally not being seen in clinics at those times."

HopkinsPD works on Android, and is available at the Parkinson's Voice Initiative website. An iOS version, mPower, is available through the App Store.

A paper on the research was recently published in the journal JAMA Neurology.