Medical Monitoring of Older Adults Patients After Discharged From Heart Failure Hospitalization

Heart failure is a condition with a poor prognosis, especially in older adults with a difficult management. Overall, the mortality rate increases with age, rising by 27% per decade in men and 61% per decade in women.

The management and monitoring of heart failure rely on the appropriate prescription of drugs with beneficial effect on heart muscle remodeling (ACE inhibitors, beta-blockers), with a regular reevaluation for a best efficacy as well as a good bservance of hygienic-dietary rules such as regular weight control, simple nutritional rules (salt and water intake), and the maintenance of physical activity and training.

The main aim of ambulatory services is to detect episodes of decompensation early to treat it rapidly and prevent hospitalization. In the absence of improvement, hospitalization is necessary. Once discharged from the health care facility, coordination with the town or host structure is essential to ensure appropriate continuity of care.

Poor publications have already shown the benefits of Machine Learning (ML) for predicting hospitalization and death in heart failure patients. Predictive performance ranges from average (AUC: 0.55) to very good (AUC 0.8).

However, to work, these models require a large number of highly medicalized variables (typically over a hundred, including medication, medical history, ECG....), making them very difficult to apply in real life.

On the other hand, satisfactory predictive models (AUC>0.7) have large temporal prediction windows (1 to 3 years), making health actions difficult to implement.

To the best of our knowledge, no publication presents a short prediction window (a few weeks) based on simple models (less than 15 variables with little medicalization).

The PRESAGE Care medical device makes it possible to observe functional changes potentially heralding major medical events, and significantly improves predictions compared with conventional models.

• Caregivers fill in an easy-to-use application (less than 2 minutes), create to predict the risk of a serious event (hospitalization, loss of autonomy, etc.).
• An alert is sent to the end-users, who are the healthcare professionals, to trigger a healthcare intervention that could, in some cases, prevent the situation from deteriorating, or enable it to be managed in a non-emergency context.

This study is based on the hypothesize that the use of the PRESAGE CARE device coupled with a health intervention based on existing health networks could be associated with a lower incidence of unscheduled re-hospitalization, with no difference in mortality. This study is based on the hypothesize that the device will be well accepted by beneficiaries, their relatives and healthcare professionals (satisfaction > 80%), and that the intervention will not be associated with an increase in healthcare expenditure, as the additional costs associated with the use of PRESAGE CARE will be offset by the reduction in expenditure linked to avoided hospitalizations.