Ontology-based decision support plateform, for evaluating the clinical severity of Covid-19 patients and orienting them towards the right medical care option – ORIENT-COVID

First, we gathered and synthesized the recommendations of several learned societies for the management of Covid-19, in order to produce 5 decision trees corresponding to various situations (care at home, in hospital,...).

Then, we proposed a visual support based on decision trees, with two innovations compared to the existing ones: first, the trees are “multi-path”, which makes it possible to reduce the size of the trees; secondly, these are dynamically presented in a “fisheye” way, that is to say that, as one navigate, the different parts of the tree change size and display more or less detail, depending on the current stage and patient profile. These two innovations allow visualizing very large and detailed tree on a normal-sized screen.

Finally, we created a functional prototype and evaluated it with medical interns on different clinical cases, comparing three groups: interns without any help, interns with a paper guide, and interns equipped with our software, “Orient-COVID”.

We have designed a software web application allowing physicians to navigate through 5 decision trees for the management of Covid-19. Evaluation on fictitious patients showed that this software significantly improved clinical practices, while paper guidelines had no effect. We also offer an editing software to update the trees or create new ones for the management of other diseases.

In terms of scientific publication, the project led to a communication in an international conference on information visualization, and an article published in an international indexed scientific journal in medical informatics, IJMI.

The interest of this system is not limited to the initial orientation, but it could especially support the medical teams in the recognition of risk and in the management of the various stages of care by following established clinical algorithms, thus contributing to standardize practices and patient follow-up.

Jammal M, Saab A, Abi Khalil C, Mourad C, Tsopra R, Saikali M, Lamy JB. Impact on clinical guideline adherence of Orient-COVID, a clinical decision support system based on dynamic decision trees for COVID19 management: a randomized simulation trial with medical trainees. International Journal of Medical Informatics 2025;195:105772
doi.org/10.1016/j.ijmedinf.2024.105772

Lamy JB, Falcoff H, Dubois S, Meneton P, Tsopra R, Saab A. Simulation trials for evaluating clinical decision support systems. Studies in health technology and informatics (STC 2025)

Lamy JB, Jammal M, Saikali M, Mourad C, Abi Khalil C, Saab A. Fisheye visualization and multi-path trees for presenting clinical practice guidelines: Methods and application to Covid-19 (presentation). International Conference Information Visualisation (iV) 2023
www.lesfleursdunormal.fr/_downloads/iv2023.pdf

In Lebanon, the COVID19 pandemic was added in March 2020 to the already very complicated context of the country which follows the conjunction of several crises affecting the Lebanese state and its population.
Currently, the management of the COVID19 pandemic in Lebanon does not include interventions planned at the central, regional or even local level to assess the risks of hospitalization or clinical severity of infected people. This can hamper the referral of patients to the right level of care appropriate to their level of clinical risk, which can negatively impact access to care in a context of limited resources.
We propose to design and evaluate an automated clinical decision support system, using symbolic artificial intelligence (based on ontologies and inference rules) which, from patient data and knowledge from the literature and the recommendations validated by learned societies, could offer a relevant clinical orientation for patients tested positive for COVID19 according to the following modalities: access to hospitalization (regular or critical), care in a community health center with medical follow-up and paramedical examinations, support by mobile teams at home or via telemedicine, self-support at home with education and recommendations supported by telephone hotline if necessary.
The interest of this system is not limited to the initial orientation, but it could especially support the care teams in the recognition of risk and in the management of the different stages of care by following established clinical algorithms, thus contributing to standardize practices as well as patient follow-up. We propose to evaluate this system first "in vitro" on fictitious cases with a consensus panel, then "in vivo" in the real context and under the aforementioned modalities of use.