Our work at Panzi

Healing the whole person.Built ondata.

Muganga is digitising Panzi Hospital's holistic model of care. By connecting medical, psychosocial, legal and socioeconomic data into one secure infrastructure, we are not just modernising records we are laying the foundation for predictive healthcare in a crisis zone.

records being digitised
≈90,000
years of continuous clinical archives
27
kinds of care, one door
4
Congolese professionals
750+

The institution

A hospital founded in 1999, now a global reference.

Panzi General Referral Hospital was established in Bukavu in 1999 by Dr Denis Mukwege. Built initially for maternal health, it soon became a centre for the care of survivors of sexual violence. In 2018 Dr Mukwege received the Nobel Peace Prize “for their efforts to end the use of sexual violence as a weapon of war and armed conflict”.

The Nobel Prize

Muganga is not Panzi. We are the technical partner digitising its archive and building the tools that read from it. The data belongs to the hospital.

What the institution represents

91,000+

survivors of sexual violence cared for

199927 years of continuous records2026

since 1999Panzi Foundation

In the peer-reviewed study

women, in the peer-reviewed study1999–2015, PLOS MedicinePLOS Medicine
48,482
women treated after complicated childbirth1999–2015PLOS Medicine
37,382

The hospital today

births a year99.1% live-birth ratePanzi Foundation
3,500+
beds reserved for survivorsand specialised gynaecological carePanzi Foundation
134
Congolese professionalsdoctors, nurses, psychologists, lawyersPanzi Foundation
750+

The One-Stop Centre model

One door in. Four kinds of care.

A survivor who walks through Panzi's door is not sent from service to service. The model, described in PLOS Medicine, is structured in four pillars delivered in one place and that is exactly what makes its data so rare.

One visitone door, one file0101Surgery, treatment, follow-upSurgery, treatment, follow-up0202Trauma, supportTrauma, support0303Assistance, proceedingsAssistance, proceedings0404Independence, reintegrationIndependence, reintegration
  • 01

    Medical

    Reconstructive surgery, treatment of infection, obstetric care and clinical follow-up over time.

  • 02

    Psychosocial

    Psychological support, group therapy and community reintegration from admission, not as an afterthought.

  • 03

    Legal

    Legal assistance and accompaniment through proceedings, for those who choose to bring a case.

  • 04

    Socioeconomic

    Training, literacy and economic support because leaving healed with no means of living is not leaving healed.

The four pillars are quoted as published in the reference study.PLOS Medicine

The architecture

How four pillars become four models.

Every stream passes through the de-identification layer. There is no route from a pillar to a model that skips it not in this diagram, and not in the system.

Secure database · de-identification layerNo model ever reads a named record0101MedicalSurgery, treatment, follow-upMedicalSurgery, treatment, follow-upPredictive recoverySevere PTSD riskPredictive recoverySevere PTSD risk0202PsychosocialTrauma, supportPsychosocialTrauma, supportClinical triageHigh-risk pregnanciesClinical triageHigh-risk pregnancies0303LegalAssistance, proceedingsLegalAssistance, proceedingsLongitudinal analyticsEfficacy over timeLongitudinal analyticsEfficacy over time0404SocioeconomicIndependence, reintegrationSocioeconomicIndependence, reintegrationGeospatial modelEmerging hotspotsGeospatial modelEmerging hotspotsPILLARSSECUREMODELS

The procedure

From paper file to model, in four moves.

01 The file

Four kinds of care, one file

A survivor's file already holds all four pillars, on paper. It is captured as it is no change of practice is asked of the clinical staff.

02 Resolution

Four streams, one person

Medical, psychosocial, legal and socioeconomic entries are matched and resolved into a single longitudinal record.

03 De-identification

Identity comes off before analysis

Identifying fields are stripped and replaced with a stable pseudonym. The link back to identity stays with the hospital, never in the analytical dataset.

Predictive recovery
Triage & risk scoring
Longitudinal efficacy
Geospatial tracking
04 In use

What the models give back

A flagged risk, a pregnancy to watch, an efficacy curve, an emerging hotspot handed to clinicians who decide.

Research & AI potential

What this dataset makes possible.

None of these models is in service today. They describe what structured data allows, and the direction we are building in with the hospital, under written agreement.

  • Medical

    Predictive recovery models

    We don't wait for trauma to compound. By analysing clinical and psychological history, our models identify patients at the highest risk of severe PTSD, letting care teams deploy targeted cognitive therapy early.

  • Clinical

    Triage & risk scoring

    Anticipating complications before they happen. Machine learning evaluates maternal health data to flag high-risk pregnancies instantly and predict potential complications in complex surgical repairs.

  • Geospatial

    Mapping and deployment

    Mapping the crisis to deploy resources. By safely and securely mapping the origin points of gender-based violence, emerging conflict hotspots can be identified and mobile clinics sent where they are needed most.

  • Longitudinal

    Proving holistic care works

    We are building a secure dataset on the long-term impact of integrated interventions. That data can demonstrate, with figures, that treating the whole person lowers healthcare costs over time and succeeds at reintegrating survivors.

Status: research. None of these models informs a clinical decision today, and none will without validation and the hospital's agreement.

The product underneath

Muganga HMS our hospital management system.

None of the above exists without a layer that talks to the systems already in place. Muganga HMS is our flagship: it plugs directly into Panzi's management systems and becomes the single source everything else reads from.

  • 01

    It plugs in, it does not replace

    HMS installs alongside what exists and connects to it. Staff do not switch tools overnight, and the hospital loses nothing during the transition.

  • 02

    One record, four services

    Admission, consultation, theatre, pharmacy, laboratory and psychosocial follow-up write into the same patient record instead of six registers that do not speak to each other.

  • 03

    Built for the real conditions

    Power cuts, thin connections, stretched teams: HMS works offline and syncs when the network returns.

  • 04

    The source for everything else

    Riposte, the research models and the management dashboards all read from HMS. One truth, not four exports.

Sources

Every figure on this page, traced back.

Figures about Panzi Hospital come from the hospital itself, from peer-reviewed literature and from the Nobel committee. We do not produce them and we do not estimate them.