Exercise Prescription in Cardiac Rehabilitation Mediated by Autonomic Function

Overview

Cardiac rehabilitation (CR) is an essential secondary prevention component in the treatment of cardiovascular diseases and one of the most cost- effective clinical interventions. Exercise training (ET) in CR programs (CRP) has unequivocal benefits in the reduction of cardiovascular adverse events, by decreasing the overactivated sympathetic tone. This ET added value can be measured by variables that express autonomic control using indirect (standard) or direct (experimental) methodologies. Direct autonomic assessment (ex. Microneurography) is accurate but unusable in daily practice, whereas standard indirect autonomic assessment using clinical parameters is imprecise, resulting in underprescription to safeguard patient safety, with less benefit to the patients. In this project, we aim to apply Machine Learning models to a set of indirect and direct variables, to make a multivariate correlation analysis and so define a normalization factor for exercise prescription.

Cardiac rehabilitation (CR) has proved to be an essential secondary prevention component of the continuum in the treatment of Cardiovascular diseases (CVD), being a Class I recommendation with level of evidence A and B on the European Society of Cardiology (ESC) and American Heart Association and American College of Cardiology (AHA/ACC) Guidelines. CR is also one of the most cost-effective clinical interventions in the treatment of CVD. These diseases, namely coronary artery disease (CAD) and heart failure (HF), are associated with autonomic dysfunction, particularly an overactivation of the autonomic sympathetic system (ASS), leading to coronary vasoconstriction, myocardial remodeling, and increased basal oxygen consumption. The main component of the CR programs (CRP) is Exercise training (ET), one of the central pillars of non- pharmacological treatment in CVD, thus preventing the above- mentioned progression of deleterious effects. The role of ET in CRP has been increasingly emphasized; however, it is still not clear, among the variety of existing training programs, which is the optimal combination and type of exercise (aerobic/anaerobic or both), frequency and duration of the sessions, whose prescription should be customized considering the patient's clinical history and the pre-CRP exam results. This limitation is pointed out as a major drawback in obtaining optimized results on CRP. The absence of a methodology that can more precisely assess and hence better quantify the effect of the prescription, safely optimizing the training plan, is one of the central problems regarding CR, and will be addressed in this research proposal putting the autonomic modulation of CV system in the center of the rational to prescribe ET in CRP. The main objective of this research plan is to draw an objective and individualized protocol to prescribe ET in CRPs based on the Autonomic output. After careful ponderation, two important but different pathologies with clearly demonstrated ASS overactivation were considered: "non- ischemic HF with reduced ejection fraction" (NIHFrEF) and "CAD without HF" (CADnonHF). The following secondary objectives contribute to the achievement of this central goal, and define the majority of the associated tasks: * Evaluation of the basal sympathetic activation pattern in patients with NIHFrEF and CADnonHF (Task2); * Definition of a normalization factor regarding sympathetic activation for ET prescription purposes in the context of CRPs (Task3) * Definition of an exercise training program for CRPs with prescription guided by the autonomic response (Task 4); In Task 2 the basal activation level of the Autonomic nervous system (ANS) will be characterized for each of the two identified conditions (NIHFrEF and CADnonHF) using a sample of 30 patients for each one to ensure a good approximation of the sampling distribution of the mean using the central limit theorem, and the participants will be enrolled at the CR consultation in Centro Hospitalar de Leiria (CHL) Cardiology department. The CHL CR Unit is accreditated by the European Association of Preventive Cardiology (EAPC) since 2022. Classical indirect measures of ANS which are relatively imprecise, such as heart rate variability (HRV) and derived indexes will be obtained with a 24h Holter and a "long duration EKG", as well as the first minute HR recovery with a Cardiopulmonary Exercise Test (CPET) and also serum catecholamines in blood and urine samples. Besides these indirect measures, a direct recording of the SNS obtained by microneurography (MSNA) will be conducted, being available at the host research institution (ciTechCare). The set of variables (autonomic and its derivatives), together with the metabolic biomarkers, will allow a multivariate correlation analysis followed by the use of the Machine Learning algorithm "Principal Component analysis" (PCA) to reduce the dimensionality of the data set, and so define a normalization factor for exercise prescription purposes, which corresponds to the main goal associated with Task 3. This normalization factor will be key to establish the individual pattern of sympathetic activation, establishing the same starting point for initial prescription of exercise and an unbiased follow up of patient's performance. Regarding Task 4, training plans (aerobic/anaerobic load) will be carried out in conjunction with the levels defined for the classification model (one of the derivable of the previous Task), and will be developed by the candidate along with the Physiatrist of the CRP Team, by setting a combined (aerobic and resistance) and stratified (with various levels of intensity, frequency, stages and duration) training program. This protocol will be evaluated by the due health ethics committees (Task 1), and all legal issues regarding safety and data protection will be respected. Regarding risks and strategies to mitigate them, the main risk is related to data assessment. In that case other hospitals may be contacted to increase the number of participants. Another risk is related to task dependency. In this case, the experience of the mentors and the integration of this project in a team with expertise in CRP and familiar with artificial intelligence applications in Medicine will be determinant.