Could AI pioneer detection and assessment of Parkinson disease?

By Jules Murtha | Medically reviewed by Kristen Fuller, MD
Published October 13, 2022

Key Takeaways

  • Researchers are developing new AI that could detect and assess Parkinson disease (PD) in patients through use of nocturnal breathing signals.

  • With an AI model, patients may be able to receive information about the progression of their PD through noninvasive, at-home assessments.

  • Clinicians have come to rely on medical histories and neurological exams to identify PD in patients. An AI model could provide a digital biomarker for PD—the first of its kind.

Recognized by the WHO as the world’s fastest-growing neurological disorder, Parkinson disease (PD) is largely diagnosed based on clinical symptoms, as there are no laboratory tests or biomarkers that can diagnose it.

But that may change through an artificial intelligence (AI) mechanism being developed to identify and track PD by analyzing patients’ nocturnal breathing signals. In the meantime, doctors can intake patients’ medical histories and perform neurological exams to diagnose PD.

How AI may help with PD

As of 2020, over 1 million individuals in the United States had been diagnosed with PD, according to a Nature Medicine article.[]

A major barrier to PD management is the lack of effective, low-cost, noninvasive diagnostic biomarkers.

To address this, researchers are looking into AI as a possible solution—and the results thus far seem promising.

According to the Nature Medicine study, the AI model is designed to detect and track PD progression through nocturnal breathing signals. PD is known to affect patients’ sleep and breathing (due to impaired respiratory muscle strength and lung capacity) in its beginning stages, so the AI would aim to catch it early on.

The study featured 7,671 individuals, several public datasets, and data from various US hospitals. To track their breathing signals, some participants wore a breathing belt on their chest or abdomen. The alternative method was to transmit a low power radio signal; researchers could then analyze the signals’ reflections off participants’ bodies.

The AI model used in this study was able to detect PD with an “area-under-the-curve of 0.90 and 0.85 on held-out and external test sets.” It could also reportedly estimate both the severity and progression of PD in patients using the same measurements as the Movement Disorder Society Unified Parkinson’s Disease Rating Scale (R = 0.94, P = 3.6 × 10–25).

The overall results pointed to AI’s potential for future detection and risk assessment of PD in patients without the need for invasive, costly tests.

The authors of the article elaborated on their design and goals for this mission, writing, “An important component of the design of this model is that it learns the auxiliary task of predicting the person’s quantitative electroencephalogram (qEEG) from nocturnal breathing, which prevents the model from overfitting and helps in interpreting the output of the model.”

"Our system aims to deliver a diagnostic and progression digital biomarker that is objective, nonobtrusive, low-cost, and can be measured repeatedly in the patient’s home."

Yang, et al.

Telltale signs of PD

Further research is needed to confirm AI’s ability to accurately track the progression analysis and preclinical diagnosis of PD in a larger patient sample.

In the meantime, doctors are left with a list of symptoms, patients’ medical histories, neurological exams, and a series of tests with which to rule out other diagnoses.

According to an article published by the National Institute on Aging, there are four main physical symptoms (all of which may manifest differently from patient to patient) associated with PD:[]

  • Slow movement

  • Tremors in the hands, arms, jaws, head, and legs

  • Muscle stiffness or extended muscle contraction

  • Balance and coordination impairments that could cause patient falls

Available treatments

Although there is no known cure for PD, there are ways to manage it.

As noted in an article published by the Cleveland Clinic, medications that add or stimulate dopamine (such as levodopa) have been shown to help patients deal with PD—although with levodopa, side effects may result from long-term use.[]

Deep brain stimulation, which can intentionally scar a part of the brain that may malfunction due to PD, is another available treatment.

Other experimental treatments such as stem cell transplants, neuron-repair treatments, and gene therapies are in the works.

What this means for you

Research suggests that AI has potential to detect and assess the progression of PD in patients. By analyzing patients’ nocturnal breathing signals, an AI model could noninvasively assess PD in patients from their homes—providing the first digital biomarker for this disorder. As this tech develops, you may continue to diagnose and treat PD based on patient symptoms, medical history, and neurological exams.

Read Next: Has AI lived up to its promise for healthcare?

Share with emailShare to FacebookShare to LinkedInShare to Twitter