Name in original language
PALOMA: Inteligencia Artificial para la Detección Temprana de la Soledad No Deseada en Personas Mayores
Initiative overview
Unwanted loneliness affects older adults over 75 who live alone in Madrid and can lead to isolation, declining health, and loss of autonomy. Early detection is difficult, as many do not seek help and traditional channels—postal mail or in person visits—are not sufficient to identify those who may be at risk. To address this challenge, the Madrid City Council launched a pilot initiative using artificial intelligence to proactively contact this population group and better understand their situation through brief telephone conversations.
PALOMA uses natural language processing technology capable of maintaining conversations with a natural sounding synthetic voice and administering a short questionnaire designed by social services professionals. The AI analyzes responses to detect signs of loneliness, classify risk levels, and offer follow up calls or information about municipal services. Cases flagged as relevant are always reviewed by professional staff, ensuring safe, ethical, and tailored intervention. This combination of AI and human oversight expands municipal capacity without replacing the professional role, aligning with principles of trustworthy and human centred AI.
The project has three main goals: improving early detection of unwanted loneliness, enabling more personalized interventions, and increasing access to activities and support services. It also modernizes the relationship between citizens and the administration, showing how AI can support social services in an efficient, accessible, and respectful way. Automated calls make it possible to reach thousands of people quickly and at a cost far lower than traditional alternatives such as call centers, significantly increasing the reach and effectiveness of social policies.
PALOMA carried out 5,163 calls, completed 2,071 conversations, and enabled 646 people to request municipal follow up. Of these, 602 were referred to social programs and 236 accepted a home visit; 84% showed signs of loneliness. The system identified 44 clear high risk cases and reached individuals who would not have contacted the city through other means. AI expanded reach, accelerated detection, and optimized resources. Results were measured through call analysis, conversion rates, referrals, and professional follow up.
A key strength of PALOMA is its capacity to scale. Traditional outreach programmes are limited by staff time, whereas AI enabled calling can reach thousands of people quickly and at low cost. The pilot contacted 5,000 people in only two weeks, leading to 646 individuals requesting follow up and enabling targeted support for those experiencing loneliness. These results demonstrate the potential for broader deployment and for replication in other municipalities facing similar demographic challenges.
