AI Algorithms Detect Parkinson’s Disease Based on Verily Study’s Monitoring Movements

New analysis using machine learning and sensor data from the Parkinson’s Disease Progress Signs Initiative (PPMI), a digital health research program sponsored by Michael J Fox Foundation for Parkinson’s Disease Research (MJFF), succeeded in distinguishing between people with or not Parkinson’s diseaseaccording to a small study led by Cohen Veterans of Biological Sciences (KVB).

Specific information, such as movement data, was recorded using the Verily Study Watch, a wrist-worn device that participants use for up to 23 hours a day for several months.

“This study demonstrates the feasibility of leveraging untethered, unlabeled wearable sensor data for accurate detection of Parkinson’s disease using powerful deep learning methods,” said Lee Lancashire, PhD, the study’s principal investigator and chief information officer at CVB. press release.

Through this combination of wearable devices and [artificial intelligence]We are one step closer to monitoring individual health care-related activity, such as motor function outside the clinic, unleashing the potential for early detection and diagnosis of diseases such as Parkinson’s.”

Suggestions for reading

the study, “Deep learning for daily monitoring of Parkinson’s disease out of the clinic using wearable sensors“in the magazine sensors.

Parkinson’s disease is a progressive neurological condition characterized by motor symptoms, such as bradykinesia (bradykinesia), gait abnormalities, and tremors. Diagnosis can be difficult especially because there are no objective biomarkers for the disease. Better ways to track disease progression are also needed so that doctors can provide individualized care and treatments.

According to the researchers, sensor technology is developing rapidly. Several studies have focused on the search for digital biomarkers associated with specific movement features of Parkinson’s disease. However, data from these studies were recorded under controlled laboratory settings and do not reflect the patients’ movements in their everyday environment.

Now, a team of researchers has sought to collect data using Watch the study in realworn daily by a subset of participants in the PPMI To study and determine whether newly developed computer-generated algorithms can be used to identify people with Parkinson’s disease based on walking-like events.

The PPMI (NCT01141023) is a longitudinal, observational study of people with and without Parkinson’s disease. Its goal is to identify biomarkers associated with the risk, onset, and progression of Parkinson’s disease.

In 2018, PPMI launched a sub-study using the Verily Study Watch at locations in the United States; All subjects registered with PPMI were invited to participate.

“Thus, compared to other data types associated with the full PPMI data set, wearable data is not limited to tracking the progress of [Parkinson’s] It begins at an early, untreated stage but may start at any point along the path, the researchers wrote.

For the new analysis, the researchers extracted participant data from 11 people from the PPMI database in June 2021: seven patients with clinical diagnoses of Parkinson’s disease and four controls. Among those with Parkinson’s disease, five had genetic differences in risk LRKK2And the GetAnd the SNCA genes, and two recently diagnosed patients who remained untreated during the study.

Patients were asked to wear the Verily Study Watch for up to 23 hours a day from several months to two years during their daily activities.

100% accuracy

The new algorithm demonstrated 100% accuracy for diagnosing Parkinson’s disease based on data from participants’ accumulated walking movements over the course of a day. According to the researchers, this can be interpreted “as the ability to identify subtle changes related to gait [Parkinson’s] whose fate does not count in the UPDRS [Unified Parkinson’s Disease Rating Scale] degrees.” UPDRS is a commonly used scale Assessment of symptom severity in Parkinson’s disease.

It can also distinguish with nearly 90% accuracy between people with a diagnosis of Parkinson’s disease and those without Parkinson’s disease in single, five-second walking-like movements.

“Although additional studies are needed, we are excited that sensor data obtained from patients’ normal activity can be used to enable clinicians to monitor and classify [Parkinson’s] “Symptoms are easily obtained through the ease of obtaining objective measures that can be used to improve clinical decision-making and guide therapeutic interventions,” said Mark Fraser, PhD, co-author of the study, and MJFF’s chief scientific officer.

Funds from CVB and a grant from the MJFF supported this study.