Predict future healthcare use

Connect dated healthcare encounters with demographics, household and insurance context to model future recorded use.

The record concepts include encounter frequency, care setting, coverage status, dated encounters, care history, health-state events and age / life stage. Inspect encounter frequency and documented care settings. Review insurance and household context at relevant dates. Evaluate agreed future-use measures without inferring unmet need.

This dataset is prepared to order. Choose the city, date range and required scope; the accepted order defines the population, final fields, formats and delivery.

Onboarded city options

Nairobi (Kenya) · Lagos (Nigeria)

Custom city requests

African Major Cities · Additional cities require availability, scope and quote confirmation.

Data type

Synthetic

Connected domains

Health, Demographics, Households, Insurance

Catalogue structure

Longitudinal

Key features

Record concepts for your order-specific extract.

Who it is for

• Health-data teams
• researchers
• insurers
• ML engineers

Entities represented

Resident-level

This identifies the entities described by the catalogue. The analytical row grain, eligibility, keys and dates are agreed for your extract.

Explore later recorded care encounters using earlier care, household and coverage histories.

Record concepts for your extract

• encounter frequency
• care setting
• coverage status
• dated encounters
• care history
• health-state events
• age / life stage

The agreed specification defines the final columns, keys, relationships and observation period for your order.

Published reference applications

• 12-month healthcare-use prediction using AnyHealthcareUseNext12M
• 12-month inpatient-admission prediction using InpatientAdmissionNext12M
• Future encounter-count modeling using EncounterCountNext12M
• Longitudinal resident trajectories and temporal-generalization analysis
• Insurance, employment and household-context comparisons within the synthetic panel

Custom study design

Fix the prediction date and encounter definitions; evaluate counts and setting probabilities separately, using pre-cutoff information and documented follow-up coverage.

The published Nairobi panel has 12-month forward targets and the stated annual temporal split. Six-month forecasts or other date schedules below are proposed custom study requirements, not the published reference benchmark.

Proposed questions for your custom order

1. Can prior care, household and insurance histories predict the number and recorded settings of healthcare encounters over the next six months?
2. Which age groups and households use different types of healthcare?
3. How does healthcare use vary across household income levels?
4. How does use of care differ between people with and without insurance?
5. Which care histories are followed by more frequent chronic-care visits?
6. How does a change in household composition coincide with care use?
7. Which households have health coverage, and who remains outside it?
8. How do covered and uncovered people use different care settings?

Proposed research and analytical applications

• Prediction prototyping
• Feature exploration
• Temporal validation
• Cohort comparison

Study and modeling guidance

Published Nairobi temporal split

Keep the same resident panel in Train=2023-09-01, Validation=2024-09-01 and Test=2025-09-01. Do not randomly reshuffle these person-period records.

Historical inputs and future targets

Prior measures end at ObservationDate; targets begin strictly after that date and extend 12 months. Identifiers, split labels, future targets and DiedDuringOutcomeWindow are excluded from model inputs.

Temporal qualifications

WardId uses current ward, historical self-employment has a documented current-IsActive approximation, and out-migration censoring is not modelled. Review those limits before constructing an as-of analysis.

Custom prediction windows

A different forecast horizon, population or observation schedule is a custom study specification. Agree the dates, target rules, leakage controls and suitable evaluation before preparation.

Data generation

FCTE synthetic city engine

Data classification

Synthetic

FCTE generation methods and published city/domain checks provide source context. The order defines the final extract and the release-specific checks, documentation and limitations supplied with it.

Reference validation

These results describe the frozen Nairobi reference release ml-10-v1.2-POST_REMEDIATION_REBUILD. Validation for a prepared custom extract is defined with that order.

The corrected 30,000-row Nairobi public reference panel; the full parent registry certification is a separate scope

Reference release modeling notes

Published temporal benchmark

Keep Train=2023-09-01, Validation=2024-09-01 and Test=2025-09-01. The same 10,000 residents occur in each period; do not randomly reshuffle person-period rows.

Historical and outcome windows

Prior measures use (ObservationDate minus the stated history window, ObservationDate]. Forward targets use (ObservationDate, ObservationDate plus 12 months].

Feature and target roles

Use the documented input columns. CitizenId, CandidateKey, SplitName, future targets and DiedDuringOutcomeWindow are excluded from predictors; the death field is diagnostic only.

Temporal qualifications

WardId is current ward rather than a separate historical ward at each observation. Historical self-employment approximates current IsActive plus StartDate; out-migration censoring is not modelled.

Reference boundaries

• The balanced panel selects residents present at all three dates, excluding later births and introducing survivorship bias; it does not estimate mortality.
• WardId uses current ward, historical self-employment uses an approximation, and out-migration censoring is not modelled.
• Synthetic temporal shifts and absolute healthcare-use levels are not calibrated empirical Nairobi trends; HLT-UTILCAL-001 remains open.
• HouseholdSize is unresolved in 7.63% of released panel rows.
• Public companions do not contain the full diagnosis and chronic-condition fact sets; reconstruction of those measures is not claimed.
• The eight visit/prescription-derived measures were reconciled against supplied companions across this 30,000-row public panel. Those targeted checks do not certify the full parent dataset, reconstruct all diagnosis/chronic-condition facts or define a future custom order.

Known limitations

• Confirm encounter definitions, complete follow-up and insurance timing. Recorded use is not clinical need, and synthetic performance does not establish clinical validity.
• Synthetic records describe generated scenarios; they do not establish observed population prevalence or real-world causal effects.

Catalogue base price

$695 USD

Price basis

Base price per selected city

Availability

Prepared after order

The displayed catalogue amount is a base price per selected city. Your requested dates, population, fields, formats and usage rights determine the agreed scope and final quote. Applicable published discounts and taxes are confirmed with that quote.

Your prepared package

Your order confirms the data files, documentation, evidence, examples and release controls included in the prepared package.

• Agreed data extract
• Schema and field documentation
• Release-specific evidence and limitations
• Agreed loading or analysis examples
• Licence, manifest, checksums and delivery note

SD_RESEARCH_DATASET/
├── data/
│   ├── health_records.parquet
│   ├── citizens.parquet
│   ├── household_history.parquet
│   └── insurance_records.parquet
├── tasks/
│   ├── task_definition.yaml
│   ├── train.parquet
│   ├── validation.parquet
│   └── test.parquet
├── evaluation/
│   ├── baseline_model.py
│   ├── metrics.json
│   └── example_notebook.ipynb
├── documentation/
│   ├── README.md
│   ├── DATA_DICTIONARY.csv
│   ├── SCHEMA.md
│   └── METHODOLOGY.md
├── evidence/
│   ├── CERTIFICATION.json
│   ├── VALIDATION_REPORT.pdf
│   └── KNOWN_LIMITATIONS.md
├── MANIFEST.json
├── CHECKSUMS.sha256
└── LICENSE.pdf

Delivery process

Confirm the cities, date range, records, formats, rights and price in writing. The order-specific extract is then prepared and supplied under the agreed payment and delivery conditions.

One secure download package for the agreed release and licensed entity.

Delivery timing, access expiry, formats and prerequisites are confirmed in the accepted order.

Support, refreshes, future releases, hosting and APIs are included only when expressly ordered.

Data Licensing Agreement · 2026-10

A non-exclusive, non-transferable licence for the named customer within the accepted scope and duration.

Permitted use

Internal research, development, testing and analytics within the accepted order.

Model training

Model training, external applications, commercial distribution and access by other legal entities must be expressly addressed in the order.

Redistribution

Raw or substantially reconstructable data cannot be resold, published or redistributed without separate written authorization.

Attribution

Preserve release provenance and applicable third-party notices. Published outputs must identify the data as synthetic.