LESNAR AI Engineering company · Nairobi

Work / Health · geospatial

MediMatch

Hospitals let surplus medical stock expire while county facilities run out, and neither can see the other. MediMatch makes the gap visible and routes surplus across it, on real roads.

Prototypesynthetic inventory · no patient recordsnever deployed in a health system

From the globe to the street

Four frames from one recording of the running product, on synthetic inventory.

INTRO01
MediMatch intro: a hand flicks a dark globe
The globethe product's own hand flicks the globe toward Africa
INTRO02
MediMatch intro: Kenya highlighted on the globe as the tap lands
Kenyathe globe settles and the tap lands on Kenya
NATIONAL03
MediMatch national view with transfers drawn along roads, Nairobi to Mandera highlighted
On the roadsNairobi to Mandera selected; every transfer follows the road network
NAIROBI COUNTY04
MediMatch Nairobi County close-up: transfers on city roads
Street scale43 transfers inside the county, ranked by distance, urgency and fit

Every line
is a road

Transfers move by truck, not by air, so every route is drawn along the road network. The routes are computed once, with our own instance of the OSRM routing engine on OpenStreetMap, and stored with the product. Nairobi to Mandera is 1,141 km on the route the router chose, via Thika, Isiolo and Marsabit: about 18 hours of driving. The router picks the fastest road, not the shortest. The public OSRM demo server is not used.

Roads © OpenStreetMap contributors · ODbL · 55 transfers on 54 stored routes

Nairobi County: transfers inside the county 43 transfers from Nairobi hubs to facilities in the county, on 43 stored road routes, 1 to 26 km each, median 10 km.
Nairobi County at street scale. The squares are the seven hubs; each dot is a facility that receives stock.
MediMatch transfers on Kenyan roads 55 transfers on stored road routes. 12 cross a county line, the longest Nairobi to Mandera at 1,141 km. The inset shows Nairobi County, where the other 43 transfers run, 1 to 26 km each. MANDERAWAJIRKILIFIMARSABITGARISSATURKANAWEST POKOTNAROKHOMA BAYMACHAKOS 1,141 KM BY ROAD

Road km, every transfer

Road distance of each MediMatch transfer, longest first. 12 cross a county line; the other 43 stay inside Nairobi County.
TransferBarRoad distance
Nairobi → Manderainsulin1,141 km
Nairobi → Wajiroxygen concentrators678 km
Nairobi → Kilifisurgical masks568 km
Nairobi → Marsabitportable ventilator522 km
Nyeri → GarissaIV fluids404 km
Uasin Gishu → Turkanaoxygen concentrators370 km
Uasin Gishu → Turkanasurgical kits370 km
Nairobi → Garissablood bags368 km
Nakuru → West Pokotvaccine cold-chain carriers264 km
Nairobi → Narokantimalarials138 km
Kisumu → Homa Bayamoxicillin107 km
Nairobi → MachakosIV fluids60 km
Within Nairobi County43 transfers, median 10 km1–26 km

The product, moving

The whole journey in one take: globe, Kenya, the roads, then Nairobi. It plays only if you ask it to.

MEDIMATCH · SCREEN RECORDING18 s
Globe to street, recordedrunning product · synthetic inventory · captured frame by frame on a virtual clock · one scripted tap, shown as a ring · the Nairobi panel is cropped above its survey strip

What is real, and what is not

  1. RealFacility names and locations are public and real. Road geometry is real, from OpenStreetMap.
  2. SyntheticEvery inventory quantity is demonstration data. No patient record is used anywhere.
  3. CuratedThe coordinator briefs on screen are prepared text, not live model output.
  4. Not researchA survey form behind earlier material showed 64 responses: 52 were generated by our synthetic-survey tool, and the provenance of the other 12 is unverified. No survey figure is presented here as a finding.

How it is built

  1. InterfaceReact 18 and TypeScript, MapLibre for the maps, three.js for the globe.
  2. EngineExpress and Zod: listings normalised into offers and urgent requests, matched urgent-first.
  3. RoutingStored road routes from a self-hosted OSRM; a labelled estimate only if a route is missing.
  4. OriginA research project at USIU-Africa, presented at a health conference hosted by Unique Conferences Canada. Source: github.com/lesnargitonga/MediMatch.

What this page does not show

  • Deployment inside any health system. It has not been deployed.
  • Real stock levels: inventory is synthetic.
  • Survey findings: the survey is not participant research.
  • Live AI: the briefs are curated text.