Predict Urinary Tract Infections

Urinary tract infections (UTIs) are the most common infections treated in outpatient settings in the United States (1). UTIs are also the fifth most common type of hospital-acquired infection, with an estimated 62,000 infections occurring annually in acute care hospitals (2). Each year, UTIs are responsible for approximately 400,000 hospitalizations, an estimated seven million office visits, and one million ED visits, resulting in roughly $10 billion in expenditures related to UTI care (3).

Size of the Problem

• $10 billion is the estimated cost associated with UTI care in the U.S. annually (3).
• 13,000 deaths are associated with healthcare-acquired UTIs each year (2).
• 69% of CAUTIs are considered to be avoidable (6).

UTIs can lead to serious complications, especially for older adults, and may also be difficult to distinguish from asymptomatic bacteriuria (ASB). Catheter-associated urinary tract infections (CAUTIs) can lead to prolonged length of stay, sepsis, and increased costs and mortality. Annually, more than 13,000 deaths are associated with healthcare-acquired UTIs (2). Differentiating UTIs from ASB in older adults is challenging, as ASB is estimated to be found in up to 16% of women older than 65 (4). In long-term care facilities (LTCs), ASB prevalence may be as high as 50% (4). ASB also remains a common reason antibiotics are prescribed, but studies have indicated that up to 75% of antimicrobial use is inappropriate (4). This potential overutilization of antibiotics can lead to serious complications (e.g. C. difficile infection).

1. Predictive Analysis of Clinical Symptoms: Utilizing artificial intelligence to analyze and predict the onset of UTIs based on the assessment of specific clinical symptoms such as dysuria, hematuria, urgency, and frequency. This allows for early identification of at-risk patients, facilitating preventive interventions before the condition worsens.
2. Continuous and Personalized Monitoring: Implement AI solutions that allow for continuous monitoring of patients, especially those with recurrent catheter use or a history of UTIs. This technology can alert healthcare professionals to significant changes in a patient's condition that might indicate the development of a UTI, allowing for a prompt and appropriate response.
3. Predictive Model for UTI Prevention: We have developed an artificial intelligence model that uses a balanced dataset, including detailed variables such as age, gender, history of hospitalizations, use of urinary catheters, and specific symptoms like dysuria, hematuria, urgency, and frequency. The target variable, 'UTI Occurrence', indicates whether the patient will develop a UTI or not, allowing healthcare organizations to implement proactive management and prevention strategies based on the early identification of high-risk patients.

• Medical Claims: Data extracted from health insurance medical claims with details about dates and place of service, diagnosis codes, key procedures, use of medical equipment, and provider specialties.
• Lab Test Results: Data about important risk markers from tests used for diagnosis, monitoring therapy, or screening, with details about specific results and abnormal indicators.
• Social Determinants of Health (SDoH): Geo-centric data with details about the social and environmental influences on people’s health and outcomes.