AI and Machine Learning-driven characterization models for post-COVID-19 condition: Enhancing personalized care in primary care settings

David Lerma Irureta, Fátima Méndez-López, Jesus Ibañez Ruiz, Jon Schoorlemmer, Carmen Vicente García, Rosa Magallón Botaya

Keywords: Artificial intelligence (AI), Machine learning (ML), Post-COVID-19 condition (PCC), Personalized healthcare, Primary care, Biomarkers, Predictive modeling

Background:

The post-COVID-19 condition (PCC) presents a significant challenge in primary care due to the lack of specific diagnostic biomarkers and the complexity of symptomatology. The integration of artificial intelligence (AI) and machine learning (ML) offers a promising approach to characterizing PCC based on immunological, biochemical, and cytokine markers, thus enabling personalized patient care.

Research questions:

Can AI and ML models effectively differentiate PCC patients from recovered individuals based on biomarker analysis?
How can AI-driven predictive models aid in personalized treatment strategies for PCC in primary care settings?

Method:

A cohort of 170 individuals (85 PCC patients and 85 recovered) was analyzed for 167 biomarkers, including biochemical, immunological, and cytokine profiles. Advanced ML algorithms, including multivariate logistic regression (MLR) and random forest (RF), were employed to identify key biomarkers that characterize PCC with high accuracy and precision. The final predictive model incorporated four critical biomarkers.

Results:

The AI models effectively distinguished PCC patients from recovered individuals, providing valuable insights into the underlying pathophysiological mechanisms. The ML-based characterization model demonstrated strong predictive performance, highlighting the potential for clinical implementation in primary care to support early diagnosis and tailored interventions.

Conclusions:

The integration of AI and ML in PCC characterization offers a robust framework for personalized patient care, particularly in primary healthcare settings managing chronic and hard-to-diagnose conditions. Future work should focus on validating these models in larger, more diverse cohorts and integrating them into routine clinical workflows to enhance diagnostic precision and patient management.
By leveraging advanced analytical tools, primary care practitioners can benefit from improved diagnostic accuracy, risk stratification, and individualized treatment planning, ultimately leading to better health outcomes and more efficient resource allocation. Moreover, the ability of AI models to analyze complex biomarker interactions offers a new paradigm for managing multifactorial diseases such as autoimmune disorders, metabolic syndromes, and other conditions with overlapping symptomatology.

Points for discussion:

How can AI-driven models improve early detection and stratification of post-COVID-19 patients in primary care? What are the ethical and regulatory challenges of implementing AI-based diagnostic tools in primary healthcare settings?

How does the integration of machine learning models impact clinical decision-making and resource allocation in primary care for chronic conditions?

What are the limitations of current biomarker-based AI models in distinguishing post-COVID-19 condition from other chronic diseases?

#191