Machine learning model may improve mortality prediction for community-acquired pneumonia patients

According to the results of a study published in CHEST.

The researchers conducted a retrospective derivation-validation study using a probabilistic causal network (PNC) to predict the risk of 30-day mortality in patients with CAP. The study included 4531 patients (bypass cohort) with CAP admitted to Barcelona Hospital Clinic in Spain between January 2003 and December 2016 and 1034 patients (validation cohort) admitted to La Fe University Hospital in Valencia in Spain between January 2012 and December 2018.

The researchers modified the SepsisFinder (SeF) CPN that originally predicted mortality in patients with sepsis to predict mortality in patients with CAP via machine learning (SeF-ML). They validated the SeF-ML model through comparisons with 4 other clinical CAP scoring systems, including Pneumonia Severity Index (PSI), Sequential Organ Failure Assessment (SOFA), Rapid Sequential Organ Failure (qSOFA) and CURB-65 (confusion; urea, >7 mmol/L; respiratory rate ≥30/min; systolic blood pressure,


Continue Reading

The researchers calculated the area under the curve (AUC) to determine the ability of each scoring system to predict 30-day mortality in patients with CAP.

For patients in the bypass cohort (median age, 73 years; males, 60%), the SeF-ML model demonstrated the highest AUC (0.801), followed by PSI (0.799), CURB-65 ( 0.759), SOFA (0.671) and qSOFA (0.642).

Similar results were noted among patients in the validation cohort (median age, 72 years; men, 62%). The SeF-ML model demonstrated an AUC of 0.826, indicating that its performance was not significantly different between the 2 cohorts (P =.051). Moreover, the AUC of the SeF-ML model was also significantly higher than the qSOFA (AUC, 0.729; P =.005) and CURB-65 (0.764; P = 0.03) and slightly higher than SOFA (AUC, 0.771; P =.14). In contrast to its performance in bypass cohort patients, the AUC of PSI was higher compared to the SeF-ML model in validation cohort patients (AUC, 0.830 vs 0.826), even though the AUC n was not significantly different between these 2 scoring systems (P =.92).

Limitations of the study include the lack of information on the disposition of patients after admission, as well as the different baseline variables and clinical characteristics of CAP between patients in the validation and derivation cohorts.

According to the researchers, “[these] the results need further validation in other cohorts from different settings to assess the actual clinical utility of SeF-ML in predicting CAP prognosis.

Reference

Cilloniz C, Ward L, Mogensen ML, et al. Machine learning model for predicting mortality in patients with community-acquired pneumonia: development and validation study. Chest. Posted July 15, 2022: S0012-3692(22)01243-0. doi:10.1016/j.chest.2022.07.005