Supervised Machine Learning for MDR Detection in Gram-Positive Pathogens: A Cross-Sectional Hospital-Based Dataset Analysis

Umaarasu  Thirunavukkarasu; Thirunavukkarasu  Dharmalingam; Suprraja  Subramani; Senthil  Kumar S; Loganathan  Thangavel; Arbind Kumar  Choudhary

Umaarasu Thirunavukkarasu Assistant Professor, Department of Microbiology, Swamy Vivekanandha Medical College Hospital and Research Institute, Tiruchengode, India
Thirunavukkarasu Dharmalingam Assistant Professor, Department of Microbiology, Government Mohan Kumaramangalam Medical College, Salem, India
Suprraja Subramani Assistant Professor, Faculty of Pharmacy, Karpagam Academy of Higher Education, Coimbatore, India
Senthil Kumar S Saveetha College of Physiotherapy, Saveetha Institute of Medical and Technical Sciences, SIMATS, Chennai, India
Loganathan Thangavel Saveetha College of Physiotherapy, Saveetha Institute of Medical and Technical Sciences, SIMATS, Chennai, India
Arbind Kumar Choudhary Department of Pharmacology, Government Erode Medical College and Hospital, Erode, India

Keywords: Multidrug Resistance, Enterococcus, Streptococcus, MRSA, Machine Learning, Antimicrobial Stewardship, Tertiary Care, Predictive Modelling

Abstract

Background: Multidrug-resistant (MDR) Gram-positive pathogens are a growing challenge in hospital practice due to limited treatment options and poor clinical outcomes. This study assessed MDR prevalence among major Gram-positive pathogens and evaluated supervised machine learning models for MDR prediction.
Methods: A cross-sectional hospital-based study was conducted on 458 clinical specimens from a tertiary care hospital in Salem, Tamil Nadu, India, including 393 culture-positive Gram-positive isolates. MDR was defined as resistance to three or more antimicrobial classes. Logistic regression, random forest, XGBoost, and multilayer perceptron models were developed using demographic, clinical, and microbiological variables and evaluated using accuracy, precision, recall, F1-score, and AUC-ROC.
Results: Enterococcus was the predominant isolate (67.0%), followed by Streptococcus (15.5%) and MRSA (3.3%). MDR was identified in 82.9% of Enterococcus isolates, 76.9% of Streptococcus isolates, and all MRSA isolates. MDR distribution did not differ meaningfully by age or gender. Random forest and XGBoost showed the most balanced performance, with accuracy of about 0.68-0.70, F1-scores of 0.77-0.80, and AUC-ROC values of 0.58-0.59. The multilayer perceptron achieved the highest recall (1.00) and F1-score (0.91), although with lower discrimination (AUC-ROC 0.46). Older age, culture negativity, and Enterococcus isolation were the strongest predictors of MDR.
Conclusions: Gram-positive pathogens in this setting showed a high burden of multidrug resistance. Machine learning models demonstrated moderate predictive value and may complement routine microbiological surveillance and antimicrobial stewardship efforts.

How to cite this article:
Thirunavukkarasu U, Dharmalingam T, Subramani S, Kumar S S, Thangavel L, Choudhary A K, Supervised Machine Learning for MDR Detection in Gram-Positive Pathogens: A Cross-Sectional Hospital-Based Dataset Analysis. J Commun Dis. 2026;58(1):132-144.

DOI: https://doi.org/10.24321/0019.5138.202618

References

Bhatt P, Patel A, Sahni AK, Praharaj AK, Grover N, Chaudhari CN, Das NK, Kulkarni M. Emergence of

multidrug resistant enterococci at a tertiary care centre. Med J Armed Forces India. 2015 Apr;71(2):139-44.

[PubMed] [Google Scholar]

Sangwan J, Lohan K, Mane P, Kumar M. Emerging crisis of multidrug-resistant enterococci from a rural tertiary

care hospital of North India. Indian J Health Sci Biomed Res (KLEU). 2021;14(2):260-4. [Google Scholar]

Taneja N, Sharma M. Antimicrobial resistance in the environment: the Indian scenario. Indian J Med Res.

;149(2):119-28. [PubMed] [Google Scholar]

Shah S, Rampal R, Thakkar P, Poojary S, Ladi S. The prevalence and antimicrobial susceptibility pattern

of Gram-positive pathogens: three-year study at a tertiary care hospital in Mumbai, India. J Lab Physicians.

;14(2):109-14. [PubMed] [Google Scholar]

Mamtora D, Saseedharan S, Bhalekar P, Katakdhond S. Microbiological profile and antibiotic susceptibility

pattern of Gram-positive isolates at a tertiary care hospital. J Lab Physicians. 2019;11(2):144-8. [PubMed]

[Google Scholar]