Renca Te Cuida: AI-powered follow-up for early detection of health risks

The problem

The Municipality of Renca faces a challenge common to many local governments: overburdened primary healthcare services, long waiting lists, and limited budgets to provide timely follow-up for people with chronic conditions and home-based care needs. Patients who need regular check-ins, such as older adults managing long-term conditions or people receiving care at home, may go weeks or months without contact from a healthcare professional.

These gaps increase the risk of avoidable complications. A change in a patient's condition, a missed medication, or an unplanned hospitalisation may go undetected until the situation has worsened. For healthcare teams already stretched thin, there is no practical way to maintain regular telephone contact with every patient who needs it using manual follow-up alone.

The solution

The municipality introduced "Renca Te Cuida" (Renca Cares for You), a system that uses artificial intelligence to make automated phone calls to patients with chronic conditions and those receiving home-based care.

The system was developed on Genesys Cloud, leveraging its native Artificial Intelligence capabilities, including Text-to-Speech (TTS) for voice generation, Speech-to-Text (STT) for transcription of responses, and Natural Language Understanding (NLU) for understanding and interpreting natural language. The solution is complemented by API integrations that allow information to be exchanged with other systems and automated actions to be executed based on the result of each call, such as updating records and sending alerts and notifications.

The system regularly contacts patients and asks simple questions about their health status. Based on the responses, it identifies early warning signs and generates alerts. These alerts are then referred to human healthcare teams, who assess each case and decide what action is needed. The AI does not make clinical decisions or replace professional care. Its role is to extend the reach of healthcare teams by maintaining regular contact with patients who would otherwise not receive follow-up between appointments.

The approach is designed to support earlier detection of health risks, help teams prioritise the most urgent cases, and strengthen continuity of care without requiring additional staffing for routine calls.

The impact

1. Over 7,600 automated calls were made, reaching approximately 3,000 individuals
During its pilot phase, the system conducted more than 7,600 automated calls. The effective contact rate was around 40%, meaning that roughly 3,000 individuals were successfully reached and engaged with the system.

2. More than 400 early alerts generated and managed by healthcare teams
The calls generated over 400 alerts that were referred to healthcare teams for assessment. These alerts enabled the identification of hospitalisations, transfers, and emerging health risks among patients with chronic conditions and those receiving home-based care.

3. Lower operational cost than manual follow-up
The automated system is expected to reduce operational costs compared with traditional manual telephone follow-up, enabling the municipality to maintain regular patient contact within its existing budget. Future expansion is expected to increase coverage and further strengthen early risk detection.

Lessons learned

• Acceptance of automated calls in a healthcare context requires careful handling. The initiative faced cultural and organisational challenges around the use of automated phone calls for health-related follow-up. Both patients and healthcare professionals needed reassurance that the AI was playing a complementary role rather than replacing human care. This was addressed through limited pilots, continuous human oversight, and clear communication about the system's purpose.
• Data quality and contact rates shaped the design. The effective contact rate of around 40% meant that not all patients could be reached by phone. The team refined call scripts and calibrated alert thresholds throughout the pilot to improve interaction quality and ensure that the alerts generated were clinically meaningful.
• Alert management needed clear protocols to avoid overburdening teams. With over 400 alerts generated during the pilot, the team needed to establish clear protocols for clinical prioritisation and referral so that healthcare staff could act on the most urgent cases without being overwhelmed by the volume of flags.

The project was shaped by Chile's Digital Transformation Law (Ley 21.180), which establishes requirements for digital public services. The law required the municipality to ensure that its digital services align with national standards, that resident data is properly secured, and that services are accessible through existing trusted systems rather than creating new, fragmented processes.