Medical insurance claims data can help predict the likelihood of autism in children
Medical insurance claims could do more than help pay for health issues; they could help predict them, according to new findings from a Penn State interdisciplinary research team published in BMJ Health & Care IT. Researchers have developed machine learning models that assess the links between hundreds of clinical variables, including doctor visits and health care services for seemingly unrelated medical conditions, to predict the likelihood of disorders autism spectrum in young children.
Insurance claim data, which is anonymized and widely available in marketing analytics datasets, provides comprehensive, longitudinal patient medical details. The scientific literature in the field suggests that children with autism spectrum disorders also often show higher rates of clinical symptoms, such as different types of infections, gastrointestinal problems, seizures, as well as behavioral indications. These symptoms are not a cause of autism but often manifest in autistic children, particularly at a young age, so we were inspired to synthesize medical information to quantify and predict this associated likelihood.”
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, training them to assess hundreds of variables to find correlations linked to an increased likelihood of autism spectrum disorders.
“Autism spectrum disorder is a developmental disability,” 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 for a clinician to make a diagnosis. The process is usually long and many children miss the window for early interventions – the most effective way to improve outcomes.”
One of the commonly used screening tools to help identify young children with a high likelihood of autism spectrum disorder is called the Modified Toddler Autism Checklist (M-CHAT), which is normally given during routine visits to healthy children at 18 and 24 months. Old. It consists of 20 questions focusing on behaviors related to eye contact, social interactions and certain physical milestones such as walking. Tutors respond based on their observations, but according to Chen, development varies so much at these ages that the tool can misidentify children. As a result, children are often not officially diagnosed until they are four or five years old, meaning they miss out on years of potential early interventions.
“Our new model, which quantifies the sum of identified risk factors together to inform the level of likelihood, is already comparable to -; and in some cases even slightly better than -; the existing screening tool,” Chen said. “When 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 into the screening tool for clinical purposes.
“A unique strength of this work is that this clinical informatics approach can be easily integrated into clinical workflow,” Liu said. “The prediction model could be integrated into a hospital’s electronic health record system, which is used to chart patient health, as a clinical decision support tool for flagging high-risk children. so clinicians and families can take action sooner.”
This work, funded by the National Institutes of Health, the Penn State Social Science Research Institute, and the Penn State College of Engineering, is the basis for a new $460,000 grant to Chen and Whitney Guthrie, a clinical psychologist at 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, by the National Institute of Mental Health.
They are using the new grant to analyze precisely how well combined data from hospital records and screening results predict autism diagnoses, as well as to explore other potential screening tools that could better equip clinicians to help their patients. .
“Not only is the current tool missing many children on the autism spectrum, but many children who are detected by our screening tools are experiencing long waiting lists due to our limited diagnostic capacity,” Guthrie said. “While it detects many children, the M-CHAT also has very high rates of false positives and false negatives, which means that many children with autism are missed and other children are referred for assessment. autism when they may not need it.Both problems contribute to the long wait – often months or even years – for further evaluation. The consequences for children who are missed by our current screening tools are particularly important because late diagnosis often means that children entirely miss the window for early intervention.Paediatricians need better screening tools to accurately identify all children who need autism assessment as soon as possible.”
Part of the problem is the limited number of psychologists, developmental pediatricians, and other pediatric developmental experts who can diagnose autism spectrum disorder. According to Chen, the solution may exist in industrial engineering.
“The key idea is to improve how we use resources,” Chen said. “With Dr. Guthrie’s clinical expertise and my group’s modeling capabilities, we aim to develop a tool that primary care physicians without specialized training can apply to perform reliable assessments to diagnose children as early as possible in order to to get the care they need as soon as possible.”
Other contributors to the paper include first author Yu-Hsin Chen, a graduate student pursuing her doctorate in industrial and manufacturing engineering who will also write her thesis on the grant work; and co-author Lan Kong, professor of public health sciences, Penn State College of Medicine.
Chen, YH. et al. (2022) Early Detection of Autism Spectrum Disorders in Young Children Through Machine Learning Using Medical Claims Data. BMJ Health and Care Informatics. doi.org/10.1136/bmjhci-2022-100544.