Understanding Digital Bio Markers to Predict Diseases: AI Health

Digital Bio Markers Image

In 2020, almost 10,000 adults cross 65 years each day in North America.  Almost 80% seniors above 65 years have one or more chronic disease. We call it the silver tsunami and needs desperate intervention if the US or Canadian Govt. were to stem the rising CMS and HealthCare costs which already has reached almost 30% of the total spend (eg., almost $1Trillion of the $3.5Trillion annual HealthCare spend in US is on seniors aged 65+ years)

Alzheimer’s disease is one of the most rapidly growing disease that is increasing the burden on the healthcare system of any country with an aging population. It has increased by almost 90% in the last twenty years and needs to be controlled by detecting it in the early stages itself. Prior to the clinical tests, the disease can be identified cognitively as well as non-cognitively. These methods involve motor, cognitive and sensory testing, for identifying MCIs and other diseases such as Parkinson’s and Alzheimer’s disease. Even blood tests are predicting not only this but many other diseases.

So far, the tests have not been completely able to identify neurodegenerative diseases due to errors such as internal cultural biases. Longitudinal imaging methods such as structural MRIs, PET molecular scans of beta amyloid and tau proteins are the only ways to detect the disease, but are limited to research purposes due to their high costs and invasive nature. Biosensors incorporated into everyday gadgets can help us reduce the cost and time required to identify the symptoms of such neurodegenerative disease. The article provides information on using these devices and forming a digital phenotype for the clinical data.

The data collected by these devices can be categorized into active and passive data. Active data is collected when the user is prompted to collect the data or enter a metric unit for a cognitive e-test, while passive data is collected without the user even knowing. Such as the steps tracked by a smart watch.

Using everyday technology has great advantages as they are present in a large scale and have highly sensitive sensors for data collection. They are accessible from everywhere and lower the burden on the health care system. The different sensors collect information regarding the gait, motor, speech and oculomotor moment in a individual through sensors in smart watches, tablets and other biosensors. Other behavioural symptoms such as depression and social withdrawal can also be determined using various apps available on the smartphone.

Most of the patients suffering from cognitive diseases also suffer from motor impairments. Gait speed, stride length and gait variability play an important role as identifying markers for AD. Research has shown that stance time variability is linked to the central nervous system impairment while step length variability is linked to sensory impairment. Such information is collected through biosensors on wrist watches and also through contact pressure sensors on the footwear. Further accuracy can also be achieved using geo-positioning.

Use of touch screen devices such as iPads, keyboards and even a stylus can help us collect information on fine motor control and movement. More specifically apps on the mobile devices collect data on finger tapping speed and tracing accuracy. These also include collection of information on key-strokes per minute, pauses while typing that give insights on cognitive impairment. Tapping tests are currently used for identifying traits for monitoring Parkinson’s disease.

Microphone sensors can help us record different aspects of language, vocal reaction time, semantic, acoustic and syntactic voice features. Slurred speech can give us a lot of insight about potential patients.

Light sensors provide us with information on eye movements, pupil constriction and dilation. These are related to medial temporal lobe system and the cholinergic neuronal pathways and progressive neuropathological changes within the cortex. Changes in the light intensity can lead to changes in the time taken to constrict or dilate the pupil, which is assumed to be longer in patients than others.

A simple task of reading involves cognitive and oculomotor functions. Tracking data on the number of words per fixation, skipped words and accuracy can give us insights on potential patients. These patients are known to have a higher eye blink rate as compared to healthy patients.

A hallmark of Alzheimer and Dementia (AD) patients is the disruption of the cholinergic system that causes a deficit of acetylcholine and is linked to memory loss and disorientation that is a major symptom of dementia and Alzheimer’s disease. HRV (heart rate variability) in patients is found to be lower as compared to healthy controls.

Biosensors on rings, and sensing pads on mattress can track this data accurately and give us patterns on the heartbeat. Since AD patients tend to have a disturbed and inefficient sleep at night, a disrupted circadian rhythm is observed.

Apart from these motor changes, an AD patient also shows symptoms of apathy and depression. A slow withdrawal from the social circle and decreased motivation to participate in everyday activities can link a patient to AD. Apps on smart phone and watches collect passive data on individuals depending on their ability to multitask and respond to these symptoms.

AD patients experience spatial confusion and might get lost even in familiar surroundings.  A sudden decrease in the speed as compared to the rest of the traffic or changing lanes can help us collect some valuable data on the patients.

Collecting this data digitally has already shown some positive results. The data collected from these biosensors include insight on memory, language, dexterity and executive function. The collected data can then be studied and shared with the clinical specialists who can carry out further tests that will provide detailed result and validate the presence or absence of the disease.

This data can also be used further to design innovative clinical trails that might be useful in further research. One such company that is aspiring to collect data from various sources like smartphone, wearable, GPS, IoT sensors and medical devices is LocateMotion. We aspire to use existing research of high quality to create data models to predict episodes like wandering or falls or strokes initially and then with much more data be able to help medical community predict diseases like Alzheimer’s much sooner then onset.

Leave a Comment