TCFD DEVELOPS MODEL TO SUCCESSFULLY PREDICT CHALLENGING BEHAVIORS IN AUTISM

Researchers at The Center for Discovery working in collaboration with scientists at Rensselaer Polytechnic Institute have made a significant advancement in the prediction of challenging behaviors in those on the Autism Spectrum.

Together, they developed a model with a 70% success rate in accurately predicting challenging behaviors of aggression or self-injury. “Often challenging behaviors in those with complex conditions can appear to come on suddenly and be potentially harmful to others as well as to the individual,” said TCFD President Dr. Theresa Hamlin, “this work will change the future for our educators and direct care staff. With this data, we will be able to adjust daily schedules to each of our residents and increase safety and well-being for everyone. The data in essence becomes communication we are often missing when someone is non-verbal or has limited language and cannot explain their internal state.”

The Center has a vast data base gathered over decades while developing its world-renowned HealthE6 model of care are well as residential, educational, recreational and arts programming. The teams of researchers and scientists filtered enormous amounts of data on hundreds of residents, and ultimately chose an 18-month span of information on 67 TCFD pediatric residents with histories of recent intermittent aggression and 32 residents with self-injurious behavior.  Some residents had histories of both aggression and self-injurious behavior.

“This was a multi-layered complex process starting with first identifying our pediatric population with a diagnosis of autism and then reviewing and analyzing physiological data on sleep, digestive system issues like bowel movements, and number and frequency of behaviors. We also had to filter for the most complete data sets. We needed to give the algorithms as much information as we could to build more accurate models,” said Jenny Foster, TCFD Director of Adult Psychology. This data was then combined with environmental data on allergens and weather patterns, which can regularly influence the health of complex individuals.

“The strength in using this approach and this technology is that we can us develop individual models for each person too –which we found to be extremely accurate in terms of prediction – even more so than the ‘population’ model.  Those individual models helped us recognize that for some of our people, quality of sleep can make next-day challenges more likely. For others, it may be dealing with constipation. It is really affirming to our core values at TCFD and our own HealthE6 model of care, since we work very hard to stabilize those specific factors for the people we serve,” added Foster.

“This is amazing work and a window into the future value of data analytics in helping our community,” said Dr. Hamlin, “We are so proud of our team and grateful for the partnership with RPI.  It will be so valuable for our staff moving forward to know how to adjust daily life – food, exercise, programming and other stressors so that the moments of aggression and self-injury never happen. It’s knowing the past to change the future.”