How HUCA's Geriatrics team uses LOLA, Tucuvi's clinical voice agent, to follow up COPD patients aged 85+, and why age is not a barrier to AI in care.
Since September 2025, the Geriatrics team at Hospital Universitario Central de Asturias (HUCA) has been using LOLA, Tucuvi’s clinical voice agent, to follow up elderly patients with COPD.
In this article, the team shares first-hand how they have integrated the tool into their daily practice and what they are seeing.
The project team
Behind the program is the Geriatrics team of the Geriatrics Clinical Management Area in Oviedo (HUCA, Asturias):
- Elena Valle Calonge, Consultant Geriatrician
- Laura Samaniego Vega, Consultant Geriatrician
- Susana Rodríguez Arias, Geriatric Nurse Specialist
- Lucía Ovies Menéndez, Third-year Geriatrics Resident
- Taylor Aldridge Cottingham, Second-year Geriatrics Resident
- Rocío Pérez Concellón, First-year Geriatrics Nursing Resident
The content presented below is based on an interview with the team about their experience with LOLA in clinical practice.
How does clinical voice AI follow-up work in geriatric COPD patients?
Since September 2025, elderly COPD patients followed by Geriatrics at HUCA receive an automated biweekly call from LOLA.
During the call, the system collects information on respiratory symptoms, use of rescue medication, general status and possible warning signs. From these data, it generates alerts classified by severity level and provides the team with a structured summary in the clinical dashboard.
After each call, the team reviews the alerts generated by LOLA and decides on the most appropriate action in each case.
What did clinical follow-up of COPD patients look like under the traditional model?
Before incorporating the tool, follow-up of these patients was, in the team’s own words, “limited and poorly structured”. It relied on in-person consultations scheduled every 3, 6 or 12 months, depending on each patient’s clinical stability, and a major gap remained in between: no active monitoring of any kind between visits.
The consequence was predictable.
Decompensations were detected late, often when the patient already required urgent care or hospital admission. In some cases, there wasn’t even a specific Geriatrics-led follow-up, and care only arrived once an exacerbation was already underway.
The project was designed precisely to close that gap.
How do you integrate a clinical voice agent into daily practice?
The launch, they explain, was important: before calls began, the team contacted each patient or their caregiver to explain how the program worked and to adjust call timing to their situation. From there, calls were set up on a biweekly basis.
After each call, the team reviews the alerts generated by LOLA and decides on the most appropriate action in each case.
Looking back, what stands out the most is not the numbers, but the change in model:
“This system allowed us to move from a reactive model to a proactive one, enabling early detection of clinical changes, exacerbations or new care needs, and making it possible to intervene ahead of time.”
Can patients aged 85, 90 or older really use a voice agent?
Yes. In the HUCA program, 100% of patients are aged 85 or older, and 44% are over 90. Even so, this cohort answers 76% of calls and completes 88% of the calls they answer.
The team describes the reception as “overall, very satisfactory”.
What was most striking was not so much that patients answered, but the level of trust they ended up developing with the tool. Many spoke of a sense of accompaniment and of the reassurance of knowing that someone was looking out for them and would raise the alarm if anything went wrong.
Some patients even asked whether their relatives could join the program too:
“My wife has problems similar to mine, couldn’t Lola call her too?”
Adoption was not immediate in every case, something the team openly acknowledges.
Patients of very advanced age, those in institutional care, those with hearing difficulties, or those with caregivers less familiar with this type of telephone follow-up showed greater hesitation in the early phases. Most managed to adapt progressively, although not all profiles did so to the same degree.
A minority chose to leave the program, particularly when they felt that some questions did not adequately fit their clinical situation. In these cases, the team adjusted the alert configuration, recognizing that the geriatric population requires a more precise level of personalization than younger cohorts.
Has treatment adherence improved?
The team’s answer is nuanced: yes, although with variability between patients. The overall trend was positive, both in treatment adherence and in disease awareness.
Continuous follow-up allowed them to identify specific difficulties with adherence, resolve frequent questions and reinforce both pharmacological and non-pharmacological measures: review of inhaler technique, reinforcement of correct medication intake, healthy habits and self-care education.
There is also a more subtle effect that the team highlights and that deserves attention: the call itself changes the patient’s behavior.
“Having patients respond to questions on a regular basis encouraged greater involvement in their own health.”
In other words, having to articulate symptoms and answer structured questions every two weeks becomes, in itself, a form of self-monitoring.
Should age be an exclusion criterion for AI in healthcare?
Here the team is direct: no.
Their reasoning rests on three ideas.
First, that older people are not a homogeneous group: there are significant differences in autonomy, social support and clinical needs, and that’s why tools have to be flexible and person-centered, leaving room for caregivers to take part when needed.
Second, that AI used as support and always under clinical supervision does not replace the geriatrician or the nurse, but rather extends what they can do between visits.
And third, that leaving older patients out of these innovations would only “widen the care gap in a population that, precisely, requires more follow-up”.
The team sums it up in a sentence that reframes the entire debate:
“The question is not whether older patients can benefit from AI, but how to adapt these tools to improve their quality of life, preserve their autonomy and better meet their needs.”
Five takeaways for other Geriatrics teams
Drawing on HUCA’s experience, here are five ideas that can be useful for any Geriatrics service considering bringing AI into their follow-up:
- Age is not a barrier. Patients aged 85, 90 and even 95 engage with a voice agent when the setup is adapted to their clinical reality.
- Onboarding makes the difference. The first contact with the patient or caregiver, explaining the program and adapting call timing, is what drives adoption.
- Alerts have to be designed for the geriatric patient. Standard severity rules need adjustment to reflect the variability and complexity of these patients.
- AI extends, it does not replace. The tool allows the team to follow up more patients and more closely, but clinical decisions remain with the professional.
- The real change is the shift from reactive to proactive. Detecting exacerbations and adherence problems earlier improves outcomes and prevents unnecessary admissions.
The HUCA program conveys a message that goes beyond COPD: technology does not replace clinical judgment or human contact — it frees up time and attention to direct them where they’re most needed. In Geriatrics, where each patient presents a unique reality and complexity is cumulative, this is the difference between a timely intervention and a late one. It shows, once again, that progress in care is not about choosing between AI or professionals, but about integrating them effectively to work together.
