With an ageing population comes a growing burden of age-related cognitive decline. While evidence supports a relationship between higher diet quality and better cognitive outcomes, the association between ultra-processed foods (UPFs) and cognitive function remains underexplored. We examined this association using prospective data from the ASPirin in Reducing Events in the Elderly (ASPREE) study, a cohort of Australian adults aged 70 years and older. Dietary intake was assessed via a mail-based diet screening questionnaire and categorised according to the Nova classification. Cognitive performance was assessed using the Controlled Oral Word Association Test (COWAT), Symbol Digit Modalities Test (SDMT), Hopkins Verbal Learning Test–Revised (HVLT-R), Modified Mini-Mental State Examination (3MS), and a composite cognitive Z score. Participants were stratified into high UPF (≥ 4 servings/day) and low UPF (< 4 servings/day). Marginal structural models with inverse probability of treatment weighting estimated associations, adjusting for demographic, lifestyle, and clinical confounders. Among 11,502 participants (3505 high UPF; 7997 low UPF), over a median follow-up of 5.6 (± 2.4) years, higher UPF consumption was associated with poorer performance on the 3MS (mean difference –0.28, 95
BACKGROUND:Differences in opinion concerning the contribution of Mycoplasma genitalium to pelvic inflammatory disease (PID) has resulted in inconsistencies across global testing and treatment guidelines. We conducted a systematic review and meta-analysis to determine the association between M. genitalium and PID and M. genitalium positivity within PID cases to provide a contemporary evidence base to inform clinical practice (PROSPERO registration: CRD42022382156). METHODS:PubMed, Embase, Medline, and Web of Science were searched to 1 December 2023 for studies that assessed women for PID using established clinical criteria and used nucleic acid amplification tests to detect M. genitalium. We calculated summary estimates of the (1) association of M. genitalium with PID (pooled odds ratio [OR]) and 2) proportion of PID cases with M. genitalium detected (pooled M. genitalium positivity in PID), using random-effects meta-analyses, with 95% confidence intervals (CI). RESULTS:Nineteen studies were included: 10 estimated M. genitalium association with PID, and 19 estimated M. genitalium positivity in PID. M. genitalium infection was significantly associated with PID (pooled OR = 1.67 [95% CI: 1.24-2.24]). The pooled positivity of M. genitalium in PID was 10.3% [95% CI: 5.63-15.99]. Subgroup and meta-regression analyses showed that M. genitalium positivity in PID was highest in the Americas, in studies conducted in both inpatient and outpatient clinic settings, and in populations at high risk of sexually transmitted infections. CONCLUSIONS:M. genitalium was associated with a 67% increase in odds of PID and was detected in about 1 of 10 clinical diagnoses of PID. These data support testing women for M. genitalium at initial PID diagnosis.
Clinical Decision Support Systems (CDSS) improve patient outcomes and support sustainable health services by enhancing medical decisions. Developing rules for a CDSS is expensive due to delays in capturing and defining the rules through multiple iterations between clinicians and developers as the role of a clinician is patient care. We investigate the effectiveness of large language models (LLMs) and large reasoning models (LRMs) in generating a triaging rule set for a CDSS. We prompt various LLMs (GPT-3.5, GPT-4, GPT-4o, Gemini, Claude 3.5 Sonnet) and various LRMs (GPT-o1-mini, Grok-4, Claude 4 Sonnet) using alternative prompting techniques. We compare the LLM generated rule sets against the clinical rule set from our Pandemic Intervention Monitoring System (PiMS); a triaging CDSS built in collaboration with clinicians to monitor COVID-19 positive patients. Effectiveness is evaluated based on the accuracy, interpretability, and rule complexity. We identified that LLMs generated COVID-19 screening rule sets compared to triaging rule sets when not specifying the variables from our PiMS rule set. By including PiMS variables in our prompts, we discovered LLMs 1) had lower interpretability and rule complexity compared to the PiMS rule set, and 2) resulted in an average accuracy between 31.62
