Determination of Trichomonas vaginalis, certain bacterial and fungal positivity, and risk factors in female patients with urinary and genital complaints

This study aimed to determine the regional prevalence of Trichomonas vaginalis infection in Turkey and to evaluate the distribution of urogenital findings including T. vaginalis, bacterial vaginosis, Candida albicans, Candida glabrata, acute cystitis, coccobacilli, calcium oxalate crystals, lactobacilli, and hematuria among women presenting with urogenital complaints in and around Ordu province. In addition, sociodemographic and socioeconomic factors associated with these conditions were explored, and their predictive contribution was assessed using a machine learning–based multilayer perceptron (MLP) model. Smear samples were collected from 236 women attending the Gynecology Outpatient Clinics of Ordu University Training and Research Hospital. Diagnostic evaluation was performed using direct microscopy, Giemsa staining, culture methods, and Papanicolaou (Pap) staining. Sociodemographic data including age, marital status, education level, economic status, place of residence, employment status, spouse’s education and employment status, household composition, and level of knowledge regarding infectious diseases—were obtained using a structured questionnaire. Associations and potential risk factors were evaluated using multivariate analyses and an MLP neural network model. Overall, T. vaginalis positivity was detected in 24.6% (n = 58) of cases using direct microscopy, staining, and culture methods. In Pap-stained cervical smears, T. vaginalis was identified in 12.7% (n = 30), bacterial vaginosis in 23.7% (n = 56), Candida albicans in 13.6% (n = 32), coccobacilli in 14.0% (n = 33), and lactobacilli in 7.2% (n = 17) of samples. A statistically significant association was observed between Pap staining results and T. vaginalis positivity (p < 0.001). The MLP model achieved an accuracy of 77.1% in the classification of bacterial vaginosis. High negative predictive values were observed for several conditions, including T. vaginalis (97.1%), whereas positive predictive values were lower for some infections, such as Candida (53.1%). Variable importance analysis indicated that age was the most influential factor for T. vaginalis and Candida albicans, level of knowledge regarding infectious diseases for bacterial vaginosis, and employment status for hematuria. This study provides region-specific data on the prevalence of T. vaginalis, bacterial vaginosis, and fungal infections among women with urogenital complaints and demonstrates a substantial infection burden in this symptomatic population. Pap smear staining was found to be associated with T. vaginalis positivity and may serve as a supportive diagnostic indicator. The MLP model suggested that sociodemographic factors may contribute to infection prediction, highlighting the potential role of machine learning as a complementary analytical approach alongside conventional statistical methods. These findings may support future efforts in risk stratification and targeted public health interventions for urogenital infections.