Medical insurance claims data may help predict the likelihood of autism in kids

Health insurance claims can do more than just help pay for health problems; They could help predict them, according to new findings from a Penn State interdisciplinary research team published in BMJ Health and Nursing Informatics. Researchers developed machine learning models that assess the connections between hundreds of clinical variables, including doctor visits and health care services for seemingly unrelated conditions, to predict the likelihood of autism spectrum disorder in young children.

Insurance claims data, which is anonymized and widely distributed in marketing scan datasets, provides comprehensive longitudinal medical data about the patient. The scientific literature in this area suggests that children with autism spectrum disorders are often more likely to present with clinical symptoms, such as B. different types of infections, gastrointestinal problems, seizures and behavioral tips. These symptoms are not a cause of autism but are common in children with autism, particularly at a young age, so we were inspired to synthesize the medical information to quantify and predict the likelihood associated.”

Qiushi Chen, Corresponding Author, Assistant Professor of Industrial and Manufacturing Engineering, Penn State College of Engineering

The researchers fed the data into machine learning models and trained them to score hundreds of variables to find correlations linked to an increased likelihood of having autism spectrum disorders.

“Autism spectrum disorder is a developmental disorder,” said co-author Guodong Liu, associate professor of public health sciences, psychiatry and behavioral health, and pediatrics at Penn State College of Medicine. “It takes observation and multiple screenings before a doctor makes a diagnosis. The process is usually lengthy, and many children miss the early intervention window — the most effective way to improve outcomes.”

One of the commonly used screening tools to identify young children with an increased likelihood of having an autism spectrum disorder is the Modified Infant Autism Checklist (M-CHAT), which is typically used during routine visits to healthy children aged 18 and 24 months given is old. It consists of 20 questions focusing on behaviors related to eye contact, social interactions and some physical milestones like walking. Guardians respond based on their observations, but Chen says development varies so much at this age that the tool can misidentify children. As a result, children are often not formally diagnosed until age four or five, meaning they miss years of potential early intervention.

“Our new model, which quantifies the sum of identified risk factors together to determine the likelihood level, is already comparable to – and in some cases even slightly better than – the existing screening tool,” Chen said. “If we combine the model with the screening tool, we have a very promising approach for clinicians.”

According to Liu, it would be practically feasible to integrate the model with the screening tool for clinical use.

“A unique strength of this work is that this clinical informatics approach can be easily integrated into the clinical workflow,” said Liu. “The predictive model could be embedded in a hospital’s electronic medical record system, used to record patient health, as a clinical decision support tool to flag children at high risk so both physicians and families can take earlier action.”

This work, funded by the National Institutes of Health, the Penn State Social Science Research Institute, and Penn State College of Engineering, is the basis of a new $460,000 grant to Chen and Whitney Guthrie, clinical psychologists at the Children’s Hospital of Philadelphia Center, for Autism Research and Assistant Professor of Psychiatry and Pediatrics at the University of Pennsylvania Perelman School of Medicine, from the National Institute of Mental Health.

They’re using the new grant to analyze exactly how well the combined hospital data and screening results predict autism diagnoses, as well as to explore other potential screening tools that clinicians could better equip to help their patients.

“Not only does the current tool miss many children on the autism spectrum, but many children detected by our screening tools have long waiting lists due to our limited diagnostic capacity,” Guthrie said. “Although it recognizes many children, the M-CHAT also has very high rates of false positives and false negatives, meaning many autistic children are missed and other children are referred for an autism assessment when they don’t need one.” Both of these issues contribute to long waits — often many months or even years — for further screening. The implications for children, which are missed by our current screening tools, are particularly important because a delayed diagnosis often means children are missing the window for a completely miss early intervention. Pediatricians need better screening tools to accurately identify all children who need autism screening as early as possible.”

Part of the problem is the limited number of psychologists, developmental pediatricians, and other pediatric development professionals who can diagnose autism spectrum disorder. According to Chen, the solution may lie in industrial engineering.

“The key idea is to improve the way we use resources,” Chen said. “With the clinical expertise of Dr. Guthrie and my group’s modeling skills, we want to develop a tool that primary care physicians without special training can use to make confident judgments to diagnose children as early as possible so that they can get the care they need as quickly as possible.”

Other contributors include first author Yu-Hsin Chen, a graduate student pursuing her PhD in industrial and manufacturing engineering who will also write her dissertation on the fellowship work; and co-author Lan Kong, Professor of Public Health Sciences, Penn State College of Medicine.