BEACON

Bayesian Ebola Assessment for Congo

Clinical triage support for suspected Ebola Virus Disease

Developed with John Kamwina Kebela, lead and corresponding author of the source study

Why this platform exists

The early symptoms of Ebola Virus Disease — fever, vomiting, diarrhoea — resemble malaria and several other illnesses common in the same regions. RT-PCR remains the reference standard, but results take time and laboratory capacity is limited during an outbreak. Someone has to decide, now, who goes into isolation.

That decision costs something in both directions. Isolating a patient who does not have EVD exposes them to real infection risk and consumes scarce resources. Failing to isolate a true case endangers staff and other patients.

BEACON structures that decision. The platform applies a published, peer-reviewed Bayesian subjective model that turns what a clinician can observe at the bedside into an explicit probability — presented alongside what that probability is actually worth, and what it is not.

What the platform does

  • ·Computes an EVD probability from seven clinical and epidemiological factors
  • ·Distinguishes “absent” from “not asked” — a distinction the published model does not make
  • ·Explains every score: each factor's contribution, deterministically
  • ·Maintains a register of patients and RT-PCR outcomes
  • ·Measures the facility's own real-world performance

What it does not do

  • ·It does not diagnose, and does not replace RT-PCR
  • ·It never clears a patient: a low score does not exclude EVD
  • ·It does not replace clinical judgement
  • ·It has not been validated outside the DRC
  • ·It makes no decision on its own

The most important limitation, stated up front

In external validation across 450 patients, the model's specificity was 33.7% and its negative predictive value 36.5%. In plain terms: most patients it flagged did not have EVD, and of 100 patients scored below the cutoff, 64 had EVD anyway. That is why BEACON never shows a low result in green, and repeats the caveat on every screen.

Recognition

BEACON implements research carried out at the School of Public Health, University of Kinshasa, and within the Ebola response in the Democratic Republic of the Congo, in collaboration with the study's authors. The platform is a tool; the method, the data and the clinical expertise come from the people below.

Research team

12

  • John Kamwina Kebela

    Platform owner · Lead and corresponding author

    Epidemiology & Biostatistics, School of Public Health, University of Kinshasa

  • Junior Bulabula-Penge

    Co-author

    Medical doctor and microbiologist, MPH · DTM&H · Researcher at INRB · Lecturer at the Protestant University in Congo · Researcher at the University of Kinshasa

  • Prince Kimpanga

    Epidemiology & Biostatistics, School of Public Health, University of Kinshasa

  • Jean Nyandwe

    Epidemiology & Biostatistics, School of Public Health, University of Kinshasa

  • Jack Kokolomami

    Epidemiology & Biostatistics, School of Public Health, University of Kinshasa

  • Steve Bwira

    Management, School of Public Health, University of Kinshasa

  • Rostin Mabela

    Mathematics & Computer Science, Faculty of Science, University of Kinshasa

  • Olivier Mangapi

    Educational & Vocational Guidance, Ilebo Higher Pedagogical Institute

  • Berthe Barhayiga

    Anaesthesia & Intensive Care, University Clinics, University of Kinshasa

  • Godfroid Musema

    Epidemiology & Biostatistics, School of Public Health, University of Kinshasa

  • Bienvenu Kabasele

    National Institute of Public Health (NIPH/PHEOC)

  • Prof Sylvain Munyanga

    Thesis supervisor

    Management, School of Public Health, University of Kinshasa

Scientific leadership

2

  • Dr. Jean-Jacques Muyembe

    Pioneer virologist · Co-discoverer of the Ebola virus

  • Dr. Benoît Kebela

    Pioneer of the Ebola virus

Affiliations are those printed in the published article. Its complete author list, exactly as published, is on the Evidence page.

Scientific provenance

The model was developed from the consensus of seven clinical experts, then externally validated against 450 patients triaged during the DRC's 10th EVD outbreak (North Kivu and Ituri, DRC) — 695 days, 3,470 cases and 2,280 deaths.

This implementation reproduces all 14 of the paper's published Table 4 cases exactly — probability and combined likelihood ratio — and that check is re-run automatically on every change.

John Kamwina Kebela, Prince Kimpanga, Jean Nyandwe et al. (2026). Development of a Bayesian subjective model for predicting the clinical diagnosis of Ebola in the Democratic Republic of the Congo. Diagnostic Pathology.

doi:10.1186/s13000-026-01825-4

Responsible use and status

  • —BEACON is a research and decision-support prototype. It is not a certified medical device, has not been submitted for regulatory clearance in any jurisdiction, and is not cleared for autonomous clinical use. It supports a clinician's decision; it does not make one.
  • —Default decision cutoff: 80%, the study's optimal cut-off point. Every assessment stores the cutoff it was scored against.
  • —The platform stores identifiable data. It should only be deployed alongside a locally defined framework for confidentiality, consent and data retention.
  • —External validation was single-centre. Performance in other regions or outbreaks remains to be established.
  • —The model's parameters are used here with the involvement of the study's lead and corresponding author, who owns this platform. The article itself remains published under CC BY-NC-ND 4.0.

Scientific contact: john.kamwina@unikin.ac.cd