RT info:eu-repo/semantics/doctoralThesis
T1 Diseño, implementación y evaluación de un novedoso sistema de apoyo a la toma de decisiones clínicas basado en la inteligencia artificial para el manejo de pacientes en cuarentena domiciliaria por la COVID-19
A1 Alcoceba Herrero, Irene
A2 Universidad de Valladolid. Escuela de Doctorado
K1 Telemonitorización
K1 COVID-19
K1 Decision support systems
K1 Apoyo a la toma de decisiones
K1 Monitoring
K1 Monitorización
K1 Telemedicine
K1 Telemedicina
K1 32 Ciencias Médicas
AB Introduction: After the first wave of the COVID-19 pandemic, a high risk of clinical deterioration was observed in patients in home isolation, many of whom were admitted in advanced stages, with increased lethality. Although telemonitoring had been used in other pathologies, there was a lack of evidence on its usefulness in this context. The main objective of this thesis was to design, implement and evaluate a clinical decision support system for patients with COVID-19 in home quarantine. In addition, we aimed to evaluate the reliability of a monitoring bracelet, the ability of the system to detect severe developments early, its effect on clinical follow-up and predictors of severe alerts.Methodology: The thesis was divided into three phases: system design, reliability study and clinical trial. The designed system, based on artificial intelligence, allowed real-time visualisation of vital signs and self-reported symptoms, call logging and automatic alarm management, thus providing decision support to primary care or emergency professionals, depending on the severity. The second phase consisted of a pilot, observational, descriptive and longitudinal study to assess the reliability of a smart wristband to monitor hospitalised patients by COVID-19, correlating its measurements with those of a reference monitor. The second phase consisted of a pilot, observational, descriptive, longitudinal, descriptive study to assess the reliability of a smart wristband to monitor hospitalised patients for COVID-19 by correlating its measurements with those of a reference monitor. The third phase was a randomised, prospective, multicentre, single-blind, prospective clinical trial. Patients with COVID-19 in home isolation were randomised in a 1:1 ratio into two groups. The control group received conventional telephone follow-up by primary care along with access to a mobile application for self-reporting of symptoms. The intervention group also had portable devices for real-time telemonitoring of vital signs. Outcome variables on severe progression were assessed 30 days after diagnosis by an external, independent committee.Results: The clinical decision support system was designed and implemented. The pilot measurement reliability study analysed 4,490 measurements from 68 patients. The wristband showed acceptable reliability in correlating its measurements with those of the reference monitor for heart rate (r=0.61, p<0.001), temperature (r=0.52, p<0.001) and respiratory rate. However, it was not accurate for oxygen saturation and blood pressure, so a commercial pulse oximeter was incorporated into the system. The clinical trial enrolled 342 patients. Of these, 247 completed the study protocol (103 cases and 144 controls), with homogeneous socio-demographic characteristics. The case group received a more exhaustive follow-up, with a higher number of alerts (61,827 vs. 1,825; p<0.05). However, thanks to the automatic management of alerts using artificial intelligence, 57,657 (94.86%) were resolved autonomously, thus avoiding overloading healthcare professionals. The 2,961 (5.14%) medical alerts were managed by healthcare professionals, of which 405 were serious. Baseline factors associated with a higher likelihood of a severe alert were asthma (OR:1.74; 95% CI:1.22-2.48; p<0.05) and corticosteroid treatment (OR:2.28; 95% CI:1.24-4.2; p<0.05). No differences were observed in severe clinical course, which was benign in both groups. Overall, 166 (84%) patients were satisfied and 170 (86%) would recommend it, the case group was more satisfied.Conclusions: The clinical decision support system designed and evaluated was effective, providing real-time information on patient status, integrating self-reporting of symptoms, conventional follow-up and continuous telemonitoring. The artificial intelligence-based system allowed for more comprehensive monitoring without overburdening staff. Patients with asthma or corticosteroids were identified as risk profiles for severe alerts.
YR 2025
FD 2025
LK https://uvadoc.uva.es/handle/10324/77746
UL https://uvadoc.uva.es/handle/10324/77746
LA spa
NO Escuela de Doctorado
DS UVaDOC
RD 16-sep-2025