A Swansea University study has revolutionized how GPs find and diagnose people by revealing how machine learning can help detect ankylosing spondylitis (AS) inflammatory arthritis early.
The study, published in the open access journal PLOS ONE, was funded by UCB Pharma and Health and Care Research Wales and conducted by data analysts and researchers at the National Center for Population Health and Welfare Research (NCPHWR).
The team used machine learning techniques to profile the characteristics of people most likely to be diagnosed with AS, the second most common cause of inflammatory arthritis.
Machine learning, a type of artificial intelligence, is a method of data analysis that automates model building to improve performance and accuracy. Algorithms build models and make predictions and decisions based on sample data without being explicitly programmed.
Identify AS patients using a national data repository that enables anonymized person-based data linking across datasets using the Swansea University Medical School-based Secure Anonymised Information Linkage (SAIL) Databank and matched to patients with no record of condition diagnosis.
Data were analyzed separately for males and females, and models developed using feature/variable selection and principal component analysis constructed decision trees.
The survey results revealed that:
- In men, back pain, uveitis (inflammation of the middle layer of the eye), and nonsteroidal anti-inflammatory drug (NSAID) use before age 20 are associated with the development of AS.
- Women showed an older age of onset of symptoms compared to men with back pain and multiple analgesics.
- The test data had a prediction rate of about 70% to 80%. However, when applying the models to the general population, improving predictive value and reducing the time to diagnose AS may require multiple models to narrow the population over time, the team said. I felt
Dr. Jonathan Kennedy, Data Lab Manager and Principal Investigator at NCPHWR said: “Our study shows great potential for machine learning to help identify people with AS and better understand their diagnostic journey through the healthcare system.
“Early detection and diagnosis is essential to ensuring the best possible outcomes for patients. It helps.”
“However, machine learning is in the early stages of implementation. Developing it requires more detailed data to improve predictions and clinical utility.”
Professor Ernest Choy, NCPHWR Researcher and Head of Rheumatology and Translational Research at Cardiff University, added: “On average, it takes an AS patient eight years from symptom onset to being diagnosed and receiving treatment. Machine learning may provide a useful tool to shorten this delay.”
Professor Kieran Walshe, Director of Health and Care Research Wales, said: “It is exciting to see the cutting edge role machine learning can play in early identification of patients with conditions such as AS and the work being done at the National Center for Population Health and Wellbeing Research.
“Machine learning is still in its early stages, but it is clear that it has the potential to transform the way researchers and clinicians approach diagnosis, benefiting patients and their future health outcomes.”
Read the full story in the PLOS ONE journal.
