evidoria

← Back to browse

Good practice

Paloma — Madrid's AI Outreach System for Detecting Unwanted Loneliness in Older Residents

Spain · Madrid · See the Spain profile

Madrid's city government used an AI voice assistant to call 5,163 residents aged 75+ living alone; 646 were flagged for follow-up. A funded 'Paloma 2.0' expansion to all 65+ residents, with 337,000 planned calls, launches September 2026.

5,163
Residents called (28 Nov-8 Dec 2023)
2,937
Residents contacted
2,071
Full conversations completed
646
Flagged wanting follow-up support
85%
Reported having family/friends available
602
Classified lower risk and enrolled
378 (63%)
Reached in social-integration programme
236
Accepted home visit
143
Home visits completed
44
Referred for full in-person assessment (higher risk)
113+ hours
AI-conducted dialogue logged
€523,000
Paloma 2.0 budget
~337,000
Paloma 2.0 planned calls

Details

Maturity
Scaling
Promoter
Ayuntamiento de Madrid — Delegación de Políticas Sociales, Familia e Igualdad (Madrid Salud)
Period
November 2023 pilot; city-wide "Paloma 2.0" expansion approved, launching September 2026
Keywords
elderly care, social services, conversational AI, proactive outreach, public health

Context

Between 28 November and 8 December 2023, Madrid City Council's Department of Social Policies, Family and Equality ran a pilot called Paloma, an empathetic AI voice assistant phoning residents aged 75 and over who live alone.

Objectives

To identify unwanted loneliness among older residents living alone and flag cases for human social workers, explicitly designed as a 'first filter' rather than a decision-maker.

Activities

The assistant phoned 5,163 residents, reached 2,937, and completed full conversations with 2,071, logging more than 113 hours of AI-conducted dialogue.

Results

646 people (about one in three who answered the closing question) indicated they wanted follow-up support, and roughly one in three contacted residents acknowledged unwanted loneliness despite 85% saying they had family or friends available. Social workers classified 602 of the 646 as lower risk and enrolled them in a municipal social-integration programme, reaching 378 (63%); 236 accepted a home visit and 143 visits were completed. The remaining 44 were judged higher risk and referred for full in-person assessment.

Conclusions

On the strength of the pilot, Madrid approved 'Paloma 2.0,' a €523,000 service planning roughly 337,000 identification/follow-up calls, extending eligibility to residents 65+ and adding a four-level severity classification, scheduled to launch September 2026. No independent academic evaluation of either phase has been published, and outcome data on actual reductions in loneliness (rather than call/referral volumes) has not yet been reported.

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful