Applying Machine Learning Tools for Personalized Healthcare: Predicting Responses to Biologics in Rheumatoid Patients Through Comorbidity and Blood Test Analysis

Ahmad Alsaber¹, Adeeba Alherz², Huda Khraiss³, Ghaydaa Aldabie⁴, balqees alawadhi⁵, Jiazhu Pan⁶, Adel Alawadhi⁷, KHULOUD MOHAMMED⁸, Hoda Tarakmeh⁹, Aqeel Muhanna¹⁰, Yaser Ali¹¹ and mohammad Khudadah¹², and Kuwait Registry for Rheumatic Diseases (KRRD), ¹American University of Kuwait, Kuwait, Kuwait, ²Al-Amiri Hospital, Kuwait, Kuwait, ³Monash University, Melbourne, Queensland, Australia, ⁴ministry of health kuwait, kuwait, Kuwait, ⁵The Public Authority for Applied Education and Training (PAAET), Kuwait, Al Asimah, Kuwait, ⁶University of Strathclyde, Glasgow, Scotland, United Kingdom, ⁷Kuwait University, Kuwait, Kuwait, ⁸Farwaniya Hospital, Kuwait, Kuwait, ⁹MOH Kuwait, Kuwait, Kuwait, ¹⁰Ministry of Health, Kuwait, Al Farwaniyah, Kuwait, ¹¹Mubarek Al Kabeer Hospital, Kuwait city, Kuwait, ¹²ministery of health, kuwait, Kuwait

Meeting: ACR Convergence 2024

Keywords: American College of Rheumatology Criteria, Disease Activity, Health Assessment Questionnaire (HAQ), quality of care, rheumatoid arthritis

Session Information

Date: Saturday, November 16, 2024

Title: RA – Treatment Poster I

Session Type: Poster Session A

Session Time: 10:30AM-12:30PM

Background/Purpose: This study aims to employ various machine learning tools to predict the responses of rheumatoid arthritis (RA) patients to biologic treatments, using data from sensitive blood tests and comorbidity profiles. The goal is to optimize treatment choices, lower costs, and enhance patient outcomes. This approach customizes treatment based on detailed analyses of blood tests (e.g., WBC, Hemoglobin, LDL, HDL) and patient comorbidities, with machine learning algorithms identifying patterns predictive of biological treatment responses.

Methods: The study utilized general linear models with 10-fold cross-validation and feature selection based on blood test levels and comorbidities to predict treatment outcomes. The effectiveness of the model was validated through the area under the curve (AUC) metrics, with further gender and baseline disease activity score (DAS) adjustments.

Results: Our findings indicated significant differences in treatment responses based on gender and initial disease activity. A subset of 17 features provided a high predictive accuracy (average AUC of 0.86). Stratification by endotype and pathobiological mechanisms suggested potential improvements in treatment precision. This model, along with others tested, suggests that the integration of machine learning into clinical practice could enhance the prediction and management of fracture risks in diabetic populations significantly, improving treatment outcomes and potentially reducing healthcare costs.

Conclusion: This study confirms the variability in RA pathology and underscores the utility of machine learning in stratifying patients to optimize biological treatment strategies. Our findings support the integration of machine learning into clinical practices to advance personalized medicine in RA.

Disclosures: A. Alsaber: None; A. Alherz: AbbVie/Abbott, 6; H. Khraiss: None; G. Aldabie: None; b. alawadhi: None; J. Pan: None; A. Alawadhi: None; K. MOHAMMED: None; H. Tarakmeh: None; A. Muhanna: AbbVie/Abbott, 6; Y. Ali: None; m. Khudadah: None.

To cite this abstract in AMA style:

Alsaber A, Alherz A, Khraiss H, Aldabie G, alawadhi b, Pan J, Alawadhi A, MOHAMMED K, Tarakmeh H, Muhanna A, Ali Y, Khudadah m. Applying Machine Learning Tools for Personalized Healthcare: Predicting Responses to Biologics in Rheumatoid Patients Through Comorbidity and Blood Test Analysis [abstract]. Arthritis Rheumatol. 2024; 76 (suppl 9). https://acrabstracts.org/abstract/applying-machine-learning-tools-for-personalized-healthcare-predicting-responses-to-biologics-in-rheumatoid-patients-through-comorbidity-and-blood-test-analysis/. Accessed .

« Back to ACR Convergence 2024

ACR Meeting Abstracts - https://acrabstracts.org/abstract/applying-machine-learning-tools-for-personalized-healthcare-predicting-responses-to-biologics-in-rheumatoid-patients-through-comorbidity-and-blood-test-analysis/