← Back to the Impact Lab
Bendita IA × Caja La Araucana

Health Data Wiki

The system and its data, translated for you.

Before you build, you need to understand the terrain: how the Chilean health system works, what the law says about health data, and which sources you will work with. No lawyer-speak, no clinical jargon — in the language of builders.

Context by track
Prevention
Screening, early detection, health literacy
DEIS, public surveys, curated Lab datasets
Decompression
Risk-based triage, demand/capacity matching
Anonymized, aggregated waiting lists
Continuity
Chronic follow-up, precision medicine, autonomy
Hospital discharges, aggregated chronic data

Waiting lists — the national problem

The numbers this Lab wants to move

Chile’s system is designed to treat the sick, not to produce longevity. The waiting list is not a bug: it is a systemic gap. These are the numbers teams will build on.

Waiting hits hardest on lower incomes, older adults and the regions. It is not just a management problem — it is a structural inequity.

~2.4 million people
On the public system’s waiting lists
330,000–350,000
Waiting for surgery
400+ days average
For a specialist appointment
500+ days average
For surgery — with cases of 3 and 4 years
Cancer
Waiting turns treatable stages into metastatic ones
Chronic conditions
Cardiovascular, ophthalmology and trauma: waiting produces avoidable disability
Even those who get treated don’t always understand that the goal is not just to be cured: it is to live more years with autonomy. The longevity paradigm is missing.

The institutional map of Chilean health

Who does what — and where the measurable return is

For your solution to be adoptable you need to know who manages what. The system mixes public and private actors, and there is a concrete value chain where every avoided waiting day translates into measurable savings.

The return vein: COMPIN validates medical leave → compensation funds (like La Araucana) pay the subsidy → SUSESO oversees. Fewer waiting lists → fewer leaves → less spending → more country productivity.

Minsal
System steward: policies, programs and health prioritization
Fonasa
Public insurer — covers the large majority of the population
Health Services & hospitals
The care network that manages the waiting lists
COMPIN
Validates medical leave certificates
CCAF — Compensation funds
Pay work-incapacity subsidies (La Araucana pays on the order of a billion CLP/year)
SUSESO
Oversees social security, including medical leave and CCAFs
Every avoided waiting day is direct, measurable return — for the patient, the institutions and the country.

Responsible data at the Lab

The rules of the game — non-negotiable

The Lab works exclusively with anonymized, aggregated data, curated public sources or synthetic prospecting. Data governance is defined from day one: who owns, who authorizes, who manages.

Anonymized and aggregated only
Never identifiable patient data
Synthetic prospecting
Generated data that replicates real patterns without exposing people
Curated public sources
Datasets selected and validated by the organizing team
Re-identification is banned
Attempting to re-identify a dataset means immediate disqualification
Zero patient PII
Not in the dataset, not in the prompt, not in the demo
Governance from day one
Specific agreements define data ownership, authorization and management
Starting with anonymized, aggregated data is not a limitation: it is the condition for your solution to scale into the real system.

AI in health, done responsibly

Assist the professional — never replace them

Lab solutions extend the reach of health professionals: they let them serve more people without losing quality. What they do not do — and cannot do — is diagnose or prescribe autonomously.

Assistance, not diagnosis
Your solution supports clinical decisions; it never delivers autonomous diagnosis or medical indication
Human in the loop
Clinical decisions are made by a health professional
Cite your evidence
Every clinical claim needs a verifiable source — or say "I don’t know"
Clinical guardrails
System prompts with explicit limits on what the agent must not answer
Claude as main engine
No real Claude API calls → disqualified
Clinical judgment on the team
Every team includes at least one health professional
You are not building an artificial doctor: you are building the bridge so a doctor, a midwife or a physiotherapist can reach 10x more people.

Datasets & sources available

What you will build with

Public sources from the Chilean health system, ready to connect with Claude MCPs, plus the Lab’s curated datasets — anonymized and aggregated — published when applications open.

DEIS — Minsal
Official statistics: hospital discharges, deaths, staffing ✅
datos.gob.cl
The State’s open data portal — health category ✅
Fonasa — Statistics
Beneficiaries, services and aggregated spending ✅
BCN — Ley Fácil
Citizen-friendly explanations of current laws ✅
Curated Lab datasets
Anonymized and aggregated — published when applications open
Synthetic prospecting
Synthetic data provided by the Lab where needed
Claude API credits per participant, Claude Code, Agent SDK and MCP included. Curated datasets and their agreements are announced when applications open.