Predictive models-assisted diagnosis of AIDS-associated Pneumocystis jirovecii pneumonia in the emergency room, based on clinical, laboratory, and radiological data

Oscar José Chagas,Fabio Augusto Rodrigues Gonçalves,Priscila Paiva Nagatomo,Renata Buccheri,Vera Lucia Pereira-Chioccola,Gilda Maria Barbaro Del Negro,Gil Benard

DOI: https://doi.org/10.1038/s41598-024-61174-4

IF: 4.6

2024-05-17

Scientific Reports

Abstract:We assessed predictive models (PMs) for diagnosing Pneumocystis jirovecii pneumonia (PCP) in AIDS patients seen in the emergency room (ER), aiming to guide empirical treatment decisions. Data from suspected PCP cases among AIDS patients were gathered prospectively at a reference hospital's ER, with diagnoses later confirmed through sputum PCR analysis. We compared clinical, laboratory, and radiological data between PCP and non-PCP groups, using the Boruta algorithm to confirm significant differences. We evaluated ten PMs tailored for various ERs resource levels to diagnose PCP. Four scenarios were created, two based on X-ray findings (diffuse interstitial infiltrate) and two on CT scans ("ground-glass"), incorporating mandatory variables: lactate dehydrogenase, O2 sat , C-reactive protein, respiratory rate (> 24 bpm), and dry cough. We also assessed HIV viral load and CD4 cell count. Among the 86 patients in the study, each model considered either 6 or 8 parameters, depending on the scenario. Many models performed well, with accuracy, precision, recall, and AUC scores > 0.8. Notably, nearest neighbor and naïve Bayes excelled (scores > 0.9) in specific scenarios. Surprisingly, HIV viral load and CD4 cell count did not improve model performance. In conclusion, ER-based PMs using readily available data can significantly aid PCP treatment decisions in AIDS patients.

multidisciplinary sciences

What problem does this paper attempt to address?

The paper attempts to address the issue of how to assist in the diagnosis of Pneumocystis pneumonia (PCP) in AIDS patients in the emergency room (ER) using clinical, laboratory, and imaging data to guide empirical treatment decisions.

Specifically, the authors evaluate the ability of various predictive models (PMs) to diagnose PCP in ERs with different resource levels, aiming to improve the accuracy of physicians' treatment decisions for AIDS patients suspected of having PCP. The main objectives of the study include:

1. **Identify predictors associated with PCP**: Determine which variables are related to the diagnosis of PCP by analyzing clinical, laboratory, and imaging data.
2. **Construct predictive models**: Build multiple predictive models based on these predictors to suit ERs with different resource levels.
3. **Evaluate model performance**: Assess the performance of different models using metrics such as accuracy, precision, recall, F1 score, and AUC.

The background of the study points out that although the incidence of PCP has declined with the introduction of antiretroviral therapy (ART) and prophylactic measures, it remains one of the common opportunistic infections in AIDS patients. Due to factors such as the lack of conventional culture systems, limited specificity of clinical symptoms, low sensitivity of commonly used diagnostic methods, and the complexity of sample collection, the diagnosis of PCP remains challenging. Therefore, the researchers hope to improve the accuracy of PCP diagnosis using machine learning methods and readily available data, thereby enhancing treatment decisions.