Skip to content
Partaw Research
Menu

In development Platform program

The signal a signed read carries

Afghanistan’s outbreak surveillance counts weekly from one facility in seven and has no channel from private clinics. The platform already receives those clinics’ chest films, and a licensed radiologist signs each one. Counted weekly by province and finding, the reads become a surveillance stream the national system does not have. The feed that counts them is being built.

Week 0 1 cases 0 seen by the weekly count

Click or tap the field to start an outbreak elsewhere.

An outbreak moving through Afghanistan’s facility network: 3,750 marks, 519 of them ringed as sentinel sites — the real proportion1. Placement is schematic and the spread is a simple contact model, not a record of any real outbreak. Cases light a mark; a sentinel site flares when the outbreak reaches it, which is the only moment the weekly count learns anything. Over a run, the count sees about one case in seven.
519 of 3,750

14% of facilities report weekly

health facilities in Afghanistan were sentinel sites for the national outbreak surveillance system — the facilities whose weekly counts reach the national level. 1
17
notifiable diseases the weekly system counts. Tuberculosis is counted separately, by the national TB programme, on a quarterly register cycle. 1
About 50,000

67% of the estimated burden is counted

Worse: +3.3% on 2023

TB cases notified in 2024 against an estimated burden on the order of 75,000 a year. Roughly one case in three is never counted. 3
65,673

Better: −18% on May

acute respiratory infection and pneumonia cases reported in June 2026 alone — the largest line in the national table, and the one a chest film speaks to directly. 702,479 for the year so far. 2

The registers are paper.
The weekly count is a sample.

At the facility, disease is recorded in a register. What reaches the national level is a weekly count of seventeen conditions from the sentinel sites — one public facility in seven — and a quarterly tally from the TB programme1. Private clinics, where much of the country’s X-ray capacity sits, are outside both. More than nine million people live over an hour’s walk from any facility at all1.

A chest film is already a surveillance instrument. Read and coded, it says whether a pneumonia is present, whether the pattern looks like tuberculosis, and — with the date, the province, and an age band — where and when. Radiology reports have tracked influenza seasons and pandemic waves against public-health counts in settings that keep such records4. The platform keeps them here.

The data

What a signed read carries

The platform was built to deliver reports, and a report that can be trusted has to say when, where, for whom, and what. Those are also the four fields surveillance needs. The left column is what every study holds today; the right is what is being added so the counts can be grouped and shared.

When

The moment the film arrived and the moment a radiologist signed, on every study.

Being added — The date the film was taken, asked at intake.

Where

The sending clinic’s province and city.

Being added — Province from a controlled list, with district — so counts can be grouped, not just read.

Who

Age and sex. No name: identity never enters the imaging path.

Being added — Age bands fixed for reporting, so no row is small enough to point at a person.

What

A licensed radiologist’s signed findings and impression, in Dari or Pashto, with a critical-finding flag.

Being added — A coded conclusion confirmed by the radiologist at signature, from a fixed vocabulary of chest findings including a tuberculosis-pattern flag. Only the signed conclusion is ever counted; software scores stay internal.

The feed

The feed, as designed

Weekly. Counts by province, finding, and age band for each epidemiological week — the cadence the national outbreak system already runs on.
De-identified by construction. The feed is an aggregate table; no study-level row ever leaves the platform, and any cell too small to be anonymous is suppressed.
Shaped for the national system. Published as an aggregate dataset in the format the national health information system ingests, so it lands beside existing counts rather than in a parallel dashboard.
Shared under a data-governance agreement with a named recipient — a national surveillance programme or the implementing entity acting for it — and never sold.

Limits

What we will not claim

This is a signal, not incidence. The films come from clinics whose patients could reach a clinic and afford an X-ray. The feed complements sentinel surveillance; it does not replace it, and its counts are never presented as population rates.
Tuberculosis does not move in weeks. The fast signal is respiratory infection and pneumonia; the tuberculosis signal is a trend in case-finding yield, read over months.
Nothing is counted from software. Every count is a licensed radiologist’s signed conclusion. Computer-aided scores route films and draft text; they never enter the feed.
The feed does not exist yet. The platform stores what the feed needs, signed and timestamped, and the pieces that turn stored reads into a weekly table are being built now. Until they are running, this page describes a design.
No surveillance authority has evaluated or endorsed this program, and nothing here implies otherwise.

Partners

Who this is built for

National surveillance & TB programmes
A private-facility channel the current system does not have, delivered as counts in the format the national health information system already uses.
Implementing entities & UN agencies
A private-sector data partner for surveillance and early-warning components of country proposals — surveillance is the first priority of the current global health-security funding rounds, for which Afghanistan is eligible. 5
Health-security funders & epidemiologists
Radiology reports have been shown to track seasonal respiratory infection and pandemic waves against public-health counts. Nobody has yet done this from photographed films in a setting without a private-sector reporting channel. 4

References

  1. 1 Ahmadi et al. (2024), “Strengthening event-based surveillance (EBS): a case study from Afghanistan.” The indicator-based system gathered weekly data on 17 notifiable diseases from 519 sentinel sites of 3,750 health facilities countrywide (2021); “no established reporting mechanism exists for disease events from private facilities.” pmc.ncbi.nlm.nih.gov/articles/PMC11059675/
  2. 2 World Health Organization, Afghanistan Emergency Situation Report No. 65, June 2026, Annex B: 65,673 new ARI–pneumonia cases in June 2026; 702,479 cumulative for the year. www.emro.who.int/images/stories/afghanistan/emergency-situation-report-65-June-2026.pdf
  3. 3 KabulNow (24 March 2026), reporting WHO figures for World TB Day: more than 50,000 TB cases recorded in Afghanistan in 2024, a 3.3% increase on the previous year, with more than 875 centres providing free diagnosis and treatment. Estimated burden: World Health Organization, Global Tuberculosis Report — Afghanistan country profile (on the order of 75,000 incident cases a year). Quarterly cohort reporting is the WHO standard for national TB programmes. kabulnow.com/2026/03/who-calls-for-action-as-tuberculosis-remains-a-major-threat-in-afghanistan/
  4. 4 Heye T. et al. (2023), “Turning radiology reports into epidemiological data to track seasonal pulmonary infections and the COVID-19 pandemic,” European Radiology. 91,751 chest imaging reports over ten years; radiologically detected infection correlated with influenza-like-illness consultations (r = 0.45) and COVID-19 cases and hospitalisations (r = 0.53–0.65). pmc.ncbi.nlm.nih.gov/articles/PMC11166749/
  5. 5 The Pandemic Fund (1 April 2026), Fourth Call for Proposals: up to US$244 million for 15 high-risk, high-need countries including Afghanistan, prioritising early-warning and disease-surveillance systems, laboratories, and workforce; proposals are submitted through an approved implementing entity. www.thepandemicfund.org/news/announcement/pandemic-fund-launches-fourth-call-proposals-targeting-high-risk-high-need-countries