I led/participated in multiple studies, clinical trials, development and validation of various applied technologies, including:
1. Movement induced cognitive training. I showed the feasibility, acceptability, and initial efficacy of a novel gamified movement-induced cognitive-training platform, the Functional Brain Trainer by Intendu (Figure 1): (1) Healthy adults benefit more from combined cognitive-control training and physical activity than from the same cognitive training lacking an active physical component. (2) People suffering from stroke or traumatic brain injury reported enjoyment and satisfaction from training with the platform in clinical / community settings, managed to perform increasingly more challenging cognitive tasks within game environments, and preliminary results showed improvements in functional tasks following training (published as a conference proceeding paper; Shochat, Maoz, Stark-Inbar, et al. Proc IEEE (2017). DOI: 10.1109/ICVR.2017.8007530.).
2. Validation of smartphone deployed gait assessment app. Gait disorders and falls are common, yet they are infrequently and subjectively evaluated. Completion-time of the Timed-Up-and-Go (TUG) test is an accepted biomarker for mobility and prediction of falls risk. We validated our app version, EncephaLog-TUG, which provides gait analysis in much more detail, offering 9 additional gait biomarkers on top of the completion-time. Healthy adults conducted TUG tests while simultaneously recorded by EncephaLog and motion sensor devices used in movement labs: motion capture cameras (MCC), pressure mat; and/ or wearable sensors (Figure 2). Results show high agreement between biomarkers, suggesting that EncephaLog can provide an accurate, yet simpler, instrumented TUG platform than more complicated and expensive alternatives (published as a journal paper; Tchelet, Stark-Inbar, Yekutieli (2019). Sensors, 19(23), 5179;
https://doi.org/10.3390/s19235179(odnośnik otworzy się w nowym oknie)).
3. Smartphone deployed cognitive battery. A major project I led in Montfort, is the development and initial validation of a cognitive battery, including digital versions of existing and accepted cognitive tests to assess memory, attention, inhibition. (Figure 3). Initial results from healthy individuals (5-90 years of age) who successfully completed our smartphone version of one attentional test - Flankers, show that adults are significantly faster than children and elders in responding to stimuli. While all age groups suffer from the expected perceptual cost when a central target is “flankered” by incompatible distractors, this cost is significantly larger in children.
4. Research and clinical validation of smartphone deployed motor biomarkers for treatment optimization, clinical decision support, and rehabilitation. As part of my role in Montfort, I helped with the development, clinical testing, dissemination, and grant writing for many studies to show various implications of EncephaLog, including the following clinical case-uses: (1) Improving diagnosis and treatment in normal pressure hydrocephalus (NPH). NPH is often underdiagnosed due to the similarity between its symptoms and those of other diseases. Using EncephaLog-TUG biomarkers and AI analyses, we improve the accuracy of NPH diagnose. The platform also enables efficacy assessment of brain shunts(Figure 4A). (2) Huntington’s disease (HD) patients cope with unique gait disturbances and frequent falls. We noticed a significant difference in the anterior-posterior sway profile (repetitive forwards-backwards movements during walking) between fallers and non-fallers (Figure 4B). Moreover, among fallers, rotation time significantly correlated with the number of falls. Thus these two biomarkers are suggested as predictors for increased falls risks in HD. (3) We tested whether EncephaLog can provide an alternative score to the two most common clinical Parkinson’s scales (mUPDRS and H&Y scales). 10 EncephaLog-TUG gait biomarkers from a random sample of PD patients were fed into an artificial neural network (ANN) predictive model, to learn from the human rater scores. The ANN then significantly and strongly predicted the mUPDRS and H&Y scores for all participants (Figure 4C).
