MRI And PET Play an Important Role in Diagnosing and Predicting Alzheimer’s Disease
Neuroimaging devices, such as magnetic resonance imaging (MRIs) and positron-emission tomography (PET) scans have been harnessed by clinicians as diagnostic tools for Alzheimer’s disease (AD). Now researchers at University of Bari in Italy have developed a machine learning model, which classifies a patient’s disease status based on structural differences observed through imaging. The program is able to differentiate healthy patients from patients with AD or mild cognitive impairment (MCI); which develops into AD within 2.5-9 years.The machine learning is based on brain images obtained from 38 Alzheimer’s disease patients and 29 healthy controls. When the program was tested, it analyzed 52 healthy controls, 48 scans from MCI cases and 48 from AD patients. The experiment showed an 86% diagnostic accuracy for healthy controls to AD patients and an 84% accuracy rate for healthy vs. MCI.
(Image from gizmodo)
In another example, researchers at McGill University in Canada developed an algorithm that predicts the possibility of developing dementia using PET images. Feature values from PET images obtained from 191 MCI patients were extracted and the prediction model was constructed using machine learning. The prediction model classified another set of PET images from 273 MCI patients (43 of which were diagnosed with Alzheimer’s disease within 2 years), and it was found to be 84% accurate.
Harnessing Smartphones to Track Changes in Cognitive Function and Depression
Smartphones are now used by billions of people around the world. Due to the frequency with which we use our smartphones throughout our day-to-day life, our behavioural patterns, thoughts and preferences can be estimated through big data analysis of our usage history. This information can be collected to develop user profiles for targeted advertising. But beyond marketing, this information could also be harnessed for health purposes.
By analyzing behavioural patterns and biometric data obtained from smartphones, it is possible to estimate changes in cognitive function. Information such as movement patterns obtained from GPS history, sleep patterns estimated based on hours of inactivity, call/data transmission variations and changes in facial and pupil features can all be used to track and help diagnose fluctuations in brain function.
Estimating Cognitive Decline with A Smartphone Camera
For example, BioEye is developing a mobile app that measures a user’s gaze, pupil diameter, and blink characteristics. Research has suggested a correlation between changes in pupil dilation and declining cognitive function. This is due to the role of the locus coeruleus, one of the first brain regions affected by AD, in the regulation of pupil diameter.
The locus coeruleus is located in the brain stem and is associated with learning and memory as well as the integration of the stress response. Primarily, the locus coeruleus releases norepinephrine, a neurotransmitter known to modulate cognitive processes. In addition, release of norepinephrine, due to stimulation of the locus coeruleus, has been proposed as a neuroprotective factor against inflammation and other age-related insults. The makers of the app posit that changes in pupil dilation could reflect dysfunctional locus coeruleus activity and be used as a marker for declining cognitive function.
(Image from USC News)
(Reference: Kang, Huffer & Wheatley. 2014)
Driving and Speech Patterns As Diagnostic Tools
MyndYou, an Israeli company, calculates cognitive indexes from smartphone data to help detect early stages of MCI. The system uses GPS information to determine whether users are driving or walking based on changes in acceleration. Under driving conditions, sudden accelerations, the number of U-turns and deviations from legal speed limits are recorded and analyzed. Furthermore, the system is capable of analyzing the statistical difference between talking and not talking, the regularity of talking, concentration and emotion from the voice of call. Lastly, the MyndYou app collects information regarding heartbeat and sleep patterns. All this information is coalesced to provide therapists with insight into the day-to-day habits of their patients, which can facilitate diagnostics. Furthermore, this app will help therapists track the progression of their patients and assess treatment efficacy.
Walking Speed Could Be Used to Determine Rate of Cognitive Decline
The insurance provided Taiyo Seimei has launched an app called InfoDeliver that may help detect dementia in their policyholders. Researchers propose slowed walking speed may be a sign of dementia or MCI. Based on this hypothesis, the app tracks fluctuations in walking speed. Should a user exhibit a sharp decrease in walking speed, the app notifies the user or their family that a change has been detected. This small observation could help improve early detection of MCI.
Can An App Predict A Relapse Into Depression?
LifeGraph, another Israeli company, is using a similar premise to develop a service which looks out for signs of relapse in patients with depression. Typically, patients suffering from depression exhibit different behavioural patterns from healthy people; some may refuse to leave the house, while others prefer to frequent novel environments. Users with depression may exhibit changes in the number of phone calls or SMS messages exchanged as well as differences in sleep patterns. By recording these variations in behavioural patterns, the service detects signs of relapse early on so doctors can intervene quickly and speed the recovery process.
It should be noted this service is not robust enough to diagnose patients with depression. Instead, the app acts as a notification service by alerting hospitals that a patient’s behaviour has changed suddenly and they may require a follow-up. Similar to the apps described above, this service is a tool to aid in diagnostics and is supplemental to the treatment process.
Will Smartphones replace fMRI and PET?
Experiment Targeting 10,000 New Yorkers Will Test Efficacy of Predictions from Smartphones
The Human Project plans to log information from 10,000 new yorkers and their phones to make health predictions. First they will obtain biological information including genetic information, brain images, cortisol levels in saliva, and intestinal flora profiles. They will also incorporate socioeconomic information, communication patterns obtained through smartphones, and environmental information determined through GPS location and other databases.
Using this information, The Human Project intends to analyze aging, changes in cognitive function and the relationship between overall health and lifestyle choices. For example, the information could be used to look for trends between genetic profiles and dietary habits. By using such large-scale data it may also be possible to use a smartphone to ascertain cerebral structural features. The expectation is there may come a time when some brain assessments can be conducted using data from your smartphone.
Presently, nearly everyone carries a smartphone at all times. Behavioural data such as movement patterns, call/SMS messaging trends, and sleep patterns can all be collected through apps and sensors within these devices. Tracking changes in behavioural data obtained through smartphones could give healthcare providers an insight into a user’s cognitive state and be used for the detection of mental illness. Considering the cost and, in some areas, significant wait times for PET scans and MRIs, smartphones may be a more convenient solution. In a sense, the future of diagnostics could be in the palm of your hand.
- Nicola Amoroso, Marianna La Rocca, Stefania Bruno, Tommaso Maggipinto, Alfonso Monaco, Roberto Bellotti, Sabina Tangaro. Brain structural connectivity atrophy in Alzheimer’s disease. arXiv:1709.02369 [physics.med-ph]
- Sulantha Mathotaarachchi, Tharick A. Pascoal, Monica Shin, Andrea L. Benedet, Min Su Kang, Thomas Beaudry, Vladimir S. Fonov, Serge Gauthier, Pedro Rosa-Neto.2017. Identifying incipient dementia individuals using machine learning and amyloid imaging. Neurobiology of Aging , Volume 59, 80 – 90
- Jennifer A. Ross, Paul McGonigle & Elisabeth J Van Brockstaele. 2015. Locus coeruleus, norepinephrine and Aβ peptides in Alzheimer’s disease. Neurobiology of Stress (2) 73-84.
- Mara Mather, Carolyn W. Harley. 2016. The Locus Coeruleus: Essential for Maintaining Cognitive Function and the Aging Brain. Trends in Cognitive Science. Volume 3, 214-226.
- Mark S. Gilzenrat, Sander Nieuwenhuis, Marieke Jepma, and Jonathan D. Cohen. 2010. Pupil diameter tracks changes in control state predicted by the adaptive gain theory of locus coeruleus function. Cogn Affect Behav Neurosci. 10(2): 252–269.
- Kang OE, Huffer KE, Wheatley TP (2014) Pupil Dilation Dynamics Track Attention to High-Level Information. PLoS ONE 9(8): e102463.
- Del Campo N, Payoux P, Djilali A, Delrieu J, Hoogendijk EO, Rolland Y, Cesari M, Weiner MW, Andrieu S, Vellas B; MAPT/DSA Study Group. 2016. Relationship of regional brain β-amyloid to gait speed. Neurology. 86(1):36-43.