Treatment of urinary tract infections is improved by using a machine learning clinical decision support tool | Technology and Health

Treatment of urinary tract infections is improved by using a machine learning clinical decision support tool | Technology
Treatment of urinary tract infections is improved by using a machine learning clinical decision support tool 

Machine learning clinical decision support tool improves treatment of urinary tract infections

A new study shows that implementing a machine learning-based clinical decision support system in primary care practices significantly improved UTI treatment success.

Researchers have found that a machine learning-based clinical decision support system (CDSS) had a significant impact on treatment success and antibiotic prescribing behavior for urinary tract infections (UTIs) when implemented in primary care practices.

According to the study, which was published earlier this month in JMIR Medical Informatics, UTIs are a major health burden worldwide, but most clinical trials of UTI treatments are conducted in female patients with uncomplicated infections.  This limits scientific evidence for effective treatments of complicated UTIs, which may affect clinical decision making

The researchers further noted that machine learning CDSS methods are rarely evaluated in clinical practice, so they sought to assess how a CDSS would affect general practitioners' (GPs') antibiotic-prescribing behavior and the success of their treatments for GPs. Urinary infections.  Because the CDSS would contain data on all types of UTI patients, including those with complications, the researchers hypothesized that its use might facilitate better clinical decision-making.

READ MORE: How accurate is the AI at classifying normal breast MRI scans | Technology

To test their hypothesis, the researchers provided CDSS software designed to help GPs with treatment options for UTI patients to 36 primary care practices in the Netherlands.

The data set for the development of the CDSS was based on EHR data from patients older than 12 years who had received antibiotic treatment for at least one UTI between 2012 and 2014. Treatment success within this data set was defined as a subsequent period of 28 days in which no further treatment was necessary.  From this, classifiers were constructed to estimate the probability of success of eight commonly used antibiotics for the treatment of UTIs.

The final data set consisted of 122,203 UTIs diagnosed in 264 primary care centers.  Due to anatomical differences between male and female patients, the 15 machine learning models that make up the CDSS were divided based on gender: eight for female patients and seven for male patients.

READ MORE: How to adopt Nutrition for longevity, according to a comprehensive study | Food and Health

After the development of CDSS, physicians in the practices that participated in the study received training on how to use the software.  They used the CDSS for four months.  A control group of 29 practices was also included and provided a benchmark for evaluating the study findings.  ICU data before the study began was collected from participating practices for comparison purposes.

In practices using the CDSS, the proportion of successful treatments increased from 75 to 80 percent over the four-month implementation period.  In the control practices, the proportion remained at 76 percent.  For treatments where the software was used, the proportion of successful treatments was 83 percent.

Measuring treatment outcomes based on gender, age, comorbidities, and whether the UTI was complicated, the researchers found that use of the CDSS significantly improved outcomes for female patients and patients older than 70 years.

READ MORE: Learn about the negative health impacts of reusing cooking oil | Food and Health 

But the use of CDSS did not significantly affect the GP's antibiotic prescribing behaviour.

These findings indicate that there is potential for machine learning methods to aid in clinical decision-making and improve outcomes in primary care practices.  However, further study and validation of machine learning-based CDSS models in clinical practice is needed, the researchers said.

Source: Health IT Analytics, Direct News 99