ADHD is the most common behavioral diagnosis in early childhood, affecting around 5% of all American children. Although hyperactivity is a core symptom of ADHD, there are no objective measures that are widely used in practice settings. LemurDx is a concept for a software system for smartwatches and mobile tablets, optimized for clinical use, that uses state-of-the-art, yet relatively low-cost, sensor technology to measure hyperactivity, with the clinical goal of differentiating children with ADHD-hyperactive presentation or ADHD combined presentation from children with typical levels of activity. LemurDx will passively collect data from smartwatch sensors. Clinicians will ask a child to wear a smartwatch with the LemurDx application for one day and then collect the de-identified data. The data will be transmitted to a HIPAA-compliant secure server. Machine learning (ML) algorithms will be used to differentiate children with clinical levels of hyperactivity from those with typical levels of activity, providing clinicians with immediate results. LemurDx is superior to research-focused tools such as traditional actigraphy in a number of ways. Along with the improved precision afforded by modern smartwatch sensors (many of which are not available on actigraphy devices) and machine learning (ML) algorithms, the team is carefully designing a complete system based on the unique needs of clinicians. This includes a comprehensive system of secure HIPAA-compliant servers, automated state-of-the- art ML algorithms, pre-programmed mobile tablets and smartwatches for clinical use, a secure and de- identified data architecture, and a simple user interface that eliminates any technological burden on clinicians.
Aim 1 of the project is to develop a commercial grade prototype of LemurDx following principles of user- centered design, with feedback from focus groups including children, parents, and clinicians.
Aim 2 of the project is to test the feasibility of collecting, storing, and analyzing data from 30 children (ages 6-11) who will wear a smartwatch with LemurDx technology for one day.
Aim 3 of the project is to test and refine several ML algorithms to achieve a high level of sensitivity and specificity to accurately classify children with and without ADHD. LemurDx is significant as there is no objective and commercially available measure of hyperactivity to assist in the diagnosis of ADHD, despite a great need in settings including pediatric primary-care practices, children?s hospitals, and behavioral health clinics. The project is consistent with NICHD?s mission to ensure, ?that all children have the chance to achieve their full potential for healthy and productive lives.? Innovations include a combination of reliable, inexpensive measurement technology combined with accurate ML algorithms to consistently and objectively measure hyperactivity in a system that is designed for clinical users. We will show that LemurDx has commercialization potential by demonstrating accuracy, cost effectiveness, and user adherence, and will lay the foundation for the development of a scalable clinical delivery system.
ADHD is the most common behavioral diagnosis in early childhood affecting around 5% of all American children. Although hyperactivity is a core symptom of ADHD, there are no objective measures that are widely used in practice settings that can assist in accurate diagnostics. This project will develop a prototype for a software system for smartwatches to measure hyperactivity with objectivity and precision using state-of-the-art sensors and machine learning algorithms.
