BIGSALUD3

Objectives

By innovatively combining infrastructure services for information processing—both in terms of storage and distributed processing—with Artificial Intelligence (AI), the project aims to provide novel solutions to challenges related to precision medicine. This approach seeks to enhance the quality of life through personalized medicine, where treatments are tailored to individual patients, and hospitals can better anticipate patient needs.

Key objectives of the project include:

• Consolidation of a Methodology for Adapting Healthcare Data: Establish a common workflow between medical teams and data analysts to exchange, understand, and transform healthcare data (medical images, genomic information, clinical data, etc.) for optimized analysis using Machine Learning techniques.

• Analysis of Healthcare Data with Machine Learning Techniques: Refine Machine Learning techniques to extract discriminative features from medical images, identify relevant variables from healthcare datasets, and produce predictive models that can assist in clinical decision-making. These techniques can also be applied to other areas of interest for a hospital.

• Clinical Support Infrastructure and Software: Optimize data analysis infrastructure to meet a hospital's needs. Expand a predictive service where a physician can input patient data into an AI-based system to obtain real-time estimates, integrating these with relevant hospital services and applications to implement the Machine Learning models developed.

Below are some identified examples of interest and the potential impact AI methodologies could have on them:

• Unplanned Readmission: Between 9% and 50% of rehospitalization episodes within 30 days of discharge are preventable. Current prediction models for this issue have low predictive power, and improving them would directly impact hospital management, financial efficiency, and patient perception of healthcare quality.

• Acute Myeloid Leukemia: The cure rate for this disease after standard chemotherapy treatment is around 40-45% in young adults. However, for patients who relapse or develop treatment resistance, the survival rate drops to around 10%. Being able to predict treatment response would help specialists determine the most appropriate therapy for each patient.

• COVID-19: Over the past year, COVID-19 has affected more than 141 million people globally, claiming at least 3 million lives. The pandemic has also strained healthcare systems worldwide due to the intensive care resources needed for treating these patients. Predicting a patient’s need for intensive care through their clinical records would help hospitals optimize their available resources.

• Breast Cancer: Breast cancer is the most common cancer among women in Spain. Although mortality rates are decreasing due to medical advances, incidence rates continue to rise. Early detection systems for breast cancer require radiologists to spend extensive hours evaluating digital mammograms. Automating this evaluation would significantly reduce the time specialists need to dedicate to this task, leading to faster care in radiology departments.