With papAI Solution, we have a tool for caregivers to improve patient safety by providing accurate and efficient assistance in the management of their health.

Benoist Alexandre

Chief Engineer in charge of Clinical Engineering Development

Industry: Medical – Healthcare.

Use case: Producing a severity index.

Region: Europe.


In 2019, the hospital center of Chalon-sur-Saône started a reflection on an AI-enabled caregiver project. Based on an artificial intelligence algorithm, it aims to provide predictive patient care and reduce the mental load of caregivers by securing their practices. While the Covid-19 crisis has slowed down the implementation of this project, it has also made the need to benefit caregivers more acute.

The COVID-19 pandemic has shown that patient care is not only organized in critical care channels. “We found ourselves receiving heavy patients in medical or post-surgical units who were not used to it, with beds equipped with complete monitoring kits [non-invasive ventilation devices, monitoring monitors, etc.], an environment with which the caregivers were not familiar,” says Stéphane Kirche, Director of Innovation and Biomedical Engineering within the Saône-et-Loire – Bresse – Morvan hospital group. The GHT (Territorial Hospital Group) support institution therefore wanted to accelerate the implementation of this connected caregiver concept, of which Stéphane Kirche is one of the promoters.


A hundred users of the software on a daily basis.

      Real-time data feedback.

Improvement of the patient pathway.

physicians in place.
hospital staff.
people are cared for in the Saône-et-Loire department.

Producing a severity index

The aim is to provide caregivers with an artificial intelligence tool capable of predicting an episode of decompensation in the patient. How? By cross-referencing monitoring information collected at the patient’s bedside, in an emergency or medical department, and subsequently at the patient’s home, with data from the electronic patient record and laboratory and radiology information systems. Ultimately, it is the data of the shared medical record that could, in addition, feed the predictive algorithm, provided that the legislator authorizes access. 

This predictive index of up to six hours allows nurses and orderlies to prioritize clinical management, by changing it or by switching the patient, if necessary, to intensive care. With the increase in outpatient care, hospitalizations are increasingly affecting elderly, dependent, and poly pathological patients. This increases the mental load of caregivers. The interest of the night nurse, for example, is to be able to prioritize the visits according to predictive alerts instead of following a continuous circuit of visits, one room after another. “A predictive tool must make it possible to improve patient safety, the relevance of care and the quality of life at work of the caregiver,” says Alexandre Benoist, clinical engineer, and nurse anesthetist at the Hospital of Chalon-sur-Saône

Reduce medical errors

The concept of augmented caregiver aims to support caregivers and reduce medical errors. “Three-quarters of them are related to communication failures. Many of them could be avoided if caregivers had all the medical information of patients,” observes Stéphane Kirche before adding: “We are looking at high-reliability industries, aeronautics for example, which have managed to reduce the risk of error related to the human factor, with a rate of 106. Our goal is to achieve the same rate. The report To Err Is Human, published in 1999 by the United States Institute of Medicine, showed that medical errors were responsible for 100,000 deaths. The study Crossing the Quality Chasm of the same institute and made public in 2001 revealed that these errors were partly related to defects in the interoperability of biomedical tools.

The predictive index in practice

Here are two scenarios of the usefulness of a predictive index of worsening of a patient’s state of health: 

In the middle of the night, an operated patient begins to have weak signs of degradation (increase in heart rate, decrease in respiratory rate, decrease in oxygen saturation and blood pressure, etc.). These data, coupled with laboratory tests that reveal the onset of anemia, generate a signal sent directly into the caregiver’s pocket. The latter is thus invited to redouble vigilance for this patient whose state of health has between 70% and 80% chance of deteriorating in the next six hours. The predictive marker transmitted to the patient must make it possible to avoid the patient a stay in intensive care. “The tool tested in surgery will then be applied to other specialties,” says Stéphane Kirche. 

A patient arrives at the emergency room with recurrent epigastric pain at the deep night. Anxious, he partially communicates his medical history to the emergency doctor who has been on duty for 18 hours. By connecting to the personal medical record, it will be able to retrieve the entire medical history of the patient. “The artificial intelligence solution (from Datategy) will identify, for example, a previous identical episode, with clinical and/or biological elements that can guide the practitioner on a diagnosis of cardiac involvement. It will issue a high prediction of risk of myocardial infarction for this patient, “says Alexandre Benoist. “Like the pilot in an airliner, the algorithm accompanies the emergency doctor in a situation of fatigue and mental overload to secure his practice. The tool searches, collates and adds up weak signals and transforms them into a strong signal”.

A regional project funded by Europe

It was in June 2021 that the project began to materialize. The Chalon-sur-Saône hospital has successfully responded to the call for projects launched by the Burgundy – Franche-Comté region as part of the post-Covid economic recovery supported by the European Regional Development Fund (ERDF). Project management started in spring 2022, once the guaranteed funds amounting to €1 million were secured. “This financial support for the acquisition of material, technical and digital resources gives us the opportunity to integrate a regional network of digital innovators,” he says. A user club set up by the region allows players from different backgrounds who want to invest in digital technology to discuss their projects and to cross uses. The Regional Health Agency of Burgundy – Franche-Comté also supports the establishment in terms of paramedical resources. “At the end of September, the Chalon-sur- Saône hospital welcomed a nursing assistant and a state-certified nurse to support the care teams in the transformation of practices related to the deployment of the AI tool (Datategy) and the implementation of this new technology that involves a new way of caring for patients,” says Stéphane Kirche.


Future opening to the city

The final stage of the project in the region should be its opening to the city, in 2024, to anticipate the management in a medical service of patients followed at home and to limit the number of visits to the emergency room or critical care. Let’s take the case of remote monitoring of a patient with heart failure at home. With a connection to the artificial intelligence tool, it will be possible to detect early signs of dysfunction that may appear several days before an emergency hospitalization (weight gain, increased heart rate, etc.). The alerted care team will be able to invite the patient to go to the hospital to adjust his treatment, with a five-day hospitalization in cardiology, rather than waiting for an emergency hospitalization with a two-week stay in acute care. “This will reduce the length of stay and increase the quality of life of the patient, who will not have had a critical episode in his chronic pathology,” explains Stéphane Kirche.

Source Article from October 2022, DSIH :The #1 Hospital Information Systems Magazine



  • Improving patient flow
  • Adapt to specific caregivers’ use
  • Develop an AI-based solution quickly
Scroll to top