Impact of the Artificial Intelligence in a Telemonitoring Programme of COPD Patients With Multiple Hospitalizations

COPD Clinical Trial

Official title:

Impact of the Artificial Intelligence (Machine Learning) in a Telemonitoring Programme of COPD Patients With Multiple Hospitalizations (telEPOC)

Clinical Trial Details — Status: Recruiting

Administrative data

• NCT number: NCT04978922
• Other study ID #: PI18/01797
• Status: Recruiting
• Phase: N/A
• Start date: January 1, 2018
• Est. completion date: June 1, 2022

Study information

• Verified date: July 2021
• Source: Hospital Galdakao-Usansolo
• Contact: Cristobal Esteban, MD
• Phone: +34 94 400 7002
• Email: cristobal.est[@]gmail.com
• Is FDA regulated: No
• Study type: Interventional

Clinical Trial Summary

Given the current situation concerning healthcare, population demographics and economy, it seems required to look for new approaches in the health system. The use of new technologies must be the main factor for this change. GENERAL OBJECTIVE: To determine the impact that the application of an artificial intelligence system (Machine Learning) could have on an active telemonitoring programme of readmitted COPD patients. Particular objectives: to determine the changes in: - The use of healthcare resources. - Patients´ quality of life. - Costs. - Load of work. - Daily clinical practice. - Inflammation markers METHODS: Based on the telEPOC programme and Machine Learning developement in this project, non-randomized intervention study, with two branches: intervention (Galdakao hospital) and control (Cruces and Basurto hospital). Sample size of at least 115 patients per hospital (115 in the intervention branch and 230 in the control branch). A 2-year follow-up. Uni and multivariate statistics will be applied.

Description:

Telemonitoring programmes are an alternative to the traditional systems of patients' control, specially in chronic diseases. This kind of tools are also important because of the aging of the population, the increase in chronic diseases and the consequent increase in costs of maintenance of the health systems. On the other hand, nowadays these chronic patients are especially attended because of exacerbations, fundamentally in emergencies and hospitalization, and also in in-person scheduled consultations when patients are stable. Then, a closer attention is more desirable by the point of view of clinic, management, and costs. COPD (Chronic Obstructive Pulmonary Disease) is a highly prevalent disease. Moreover, it has a high consumption of sanitary resources and costs, 50% of whom are due to hospitalizations. Furthermore, exacerbations in COPD and specially the severe ones, have important consequences for patients (decrease of pulmonary function, worsening of quality of life and increase in mortality). Because of that, telemonitoring appears to be a solution to improve the control of these patients and improve the consumption of resources. In Galdakao Hospital in Spain, it was initiated a telemonitoring programme in COPD patients who re-admit to hospital. Its primary objective was tos reduce readmissions because of COPD exacerbations and it could demonstrate a significant decrease in the use of sanitary resources (hospitalizations, visit to emergences department, readmissions and average stay days). It also demonstrated a less worsening in clinical symptoms and quality of life in more severe patients. However, there are three factors that are very important in chronic diseases: the increase in aging people, the increase of people with chronic diseases and the fast evolution of technology, specially the recollection and information processing systems. Machine Learning (ML) is the most important part of de Artificial Intelligence, and its objective is the learning of a computer. The computer writes its own programmes to solve problems that we do not know how to solve. When works are difficult, like doing predictions in medical scenarios, ML algorithms need a high quantity of dates to get the learning. Most medical data bases have inconveniences that come from human intervention, like missing data, wrong values, etc. Because of that, programmes based on telemedicine appears to be an ideal platform for ML algorithms. This is because telemedicine systems normally produce a periodic flow of collected data by electronic ways and they are directly saved in a data base. This constant flow of dates and the low participation of people in the recollection and storage of them, give high quality to data bases, which ML algorithms can use to do the best predictions. Because of that, TelEPOC (the Telemonitoring program in a COPD cohort, in Galdakao Hospital) shows to be the best option to use in its data the ML algorithms, due to the quality and the quantity of generated data, and also because of the utility of those predictions in the clinical practice. In this situation, the question is if investigators could anticipate to an exacerbation or how much they could anticipate a manifestation of an exacerbation. To test this hypothesis, it is presented here a project that uses Artificial Intelligence (ML). Investigators previously did a test of this system, that gave promising results. That prototype was trained with retrospective data that TelEPOC programme had recollected before and it was based on an ML algorithm called Random Forests. With this probe they got a ROC curve (receiver operating characteristic curve) of 0,8 in prediction of suffering an exacerbation in following three days. Currently in Galdakao Hospital there is developing a ML system in the TelEPOC programme. Its objective is to anticipate to an alarm (exacerbation). Whit this purpose investigators consider a lot of additional questions that can be investigated, like for example: how can affect the arrival of this technology in the diary clinical practice? In this project the use of ML can change the way of focus the clinical assistance. There are tools than can predict de evolution of the patients. Another question is that if investigators anticipate an exacerbation, they could change pathogenic basis (inflammatory mediators) that round a COPD exacerbation. Investigators considerate this initiative like pioneer in this field of COPD and chronic diseases.

Recruitment information / eligibility

• Status: Recruiting
• Enrollment: 345
• Est. completion date: June 1, 2022
• Est. primary completion date: January 1, 2022
• Accepts healthy volunteers: No
• Gender: All
• Age group: 18 Years to 85 Years

Eligibility

Inclusion Criteria:

• Having a COPD (COPD was confirmed if the post-bronchodilator forced expiratory volume in one second (FEV1) divided by the forced vital capacity (FVC) was less than 0.7 (FEV1/FVC<70%)
• Having been admitted at least twice in the previous year or three times in the two previous years for a COPD exacerbation (eCOPD).

Exclusion Criteria:

• Another significant respiratory disease.
• An active neoplasm.
• A terminal clinical situation.
• Inability to carry out any of the measurements of the project.
• Unwillingness to take part in the study.

Related Conditions & MeSH terms

• COPD

Intervention

Device:
• Machine Learning: ML (Artificial Intelligence System)
To applicate an artificial intelligence system (Machine Learning: ML) on an active telemonitoring programme of readmitted COPD patients (TelEPOC)

Locations

Country	Name	City	State
Spain	Hospital Galdakao Usansolo	Galdakao	Vizcaya

Sponsors (2)

Lead Sponsor	Collaborator
Dr. Cristobal Esteban	Osakidetza

Country where clinical trial is conducted

Spain, 

References & Publications (14)

Bolton CE, Waters CS, Peirce S, Elwyn G; EPSRC and MRC Grand Challenge Team. Insufficient evidence of benefit: a systematic review of home telemonitoring for COPD. J Eval Clin Pract. 2011 Dec;17(6):1216-22. doi: 10.1111/j.1365-2753.2010.01536.x. Epub 2010 Sep 16. Review. — View Citation

Donaldson GC, Seemungal TA, Bhowmik A, Wedzicha JA. Relationship between exacerbation frequency and lung function decline in chronic obstructive pulmonary disease. Thorax. 2002 Oct;57(10):847-52. Erratum in: Thorax. 2008 Aug;63(8):753. — View Citation

Esteban C, Moraza J, Iriberri M, Aguirre U, Goiria B, Quintana JM, Aburto M, Capelastegui A. Outcomes of a telemonitoring-based program (telEPOC) in frequently hospitalized COPD patients. Int J Chron Obstruct Pulmon Dis. 2016 Nov 24;11:2919-2930. eCollection 2016. — View Citation

Esteban C, Moraza J, Sancho F et al. Machine Learning for COPD exacerbation prediction. European Respiratory Journal 2015;46:Issue suppl 59

Esteban C, Moraza J, Sancho F et al. Sistema de Alerta Temprana para el programa telEPOC mediante Machine Learning. Congreso Internacional SEPAR 2015 , Gran Canaria, España, Junio 2015.

Esteban C, Quintana JM, Moraza J, Aburto M, Egurrola M, España PP, Pérez-Izquierdo J, Aguirre U, Aizpiri S, Capelastegui A. Impact of hospitalisations for exacerbations of COPD on health-related quality of life. Respir Med. 2009 Aug;103(8):1201-8. doi: 10.1016/j.rmed.2009.02.002. Epub 2009 Mar 9. — View Citation

Esteban C, Schmidt D, Krompaß D y Tresp V. Predicting sequences of clinical events by using a personalized temporal latent embedding model. Proceedings of the IEEE International Conference on Healthcare Informatics, 2015

Jordan R, Adab P, Jolly K. Telemonitoring for patients with COPD. BMJ. 2013 Oct 17;347:f5932. doi: 10.1136/bmj.f5932. — View Citation

Mathers CD, Loncar D. Updated projections of global mortality and burden of disease, 2002-2030.World Health Organization, http//www.who.int/healthinfo/statistics/bodprojectionspaper.pdf

Noell G, Cosío BG, Faner R, Monsó E, Peces-Barba G, de Diego A, Esteban C, Gea J, Rodriguez-Roisin R, Garcia-Nuñez M, Pozo-Rodriguez F, Kalko SG, Agustí A. Multi-level differential network analysis of COPD exacerbations. Eur Respir J. 2017 Sep 27;50(3). pii: 1700075. doi: 10.1183/13993003.00075-2017. Print 2017 Sep. — View Citation

Pinnock H, Hanley J, McCloughan L, Todd A, Krishan A, Lewis S, Stoddart A, van der Pol M, MacNee W, Sheikh A, Pagliari C, McKinstry B. Effectiveness of telemonitoring integrated into existing clinical services on hospital admission for exacerbation of chronic obstructive pulmonary disease: researcher blind, multicentre, randomised controlled trial. BMJ. 2013 Oct 17;347:f6070. doi: 10.1136/bmj.f6070. — View Citation

Soler-Cataluña JJ, Martínez-García MA, Román Sánchez P, Salcedo E, Navarro M, Ochando R. Severe acute exacerbations and mortality in patients with chronic obstructive pulmonary disease. Thorax. 2005 Nov;60(11):925-31. Epub 2005 Jul 29. — View Citation

SPARRA: Scottish Patients at Risk of Readmission and Admission - A report on development work to extend the algorithm's applicability to patients of all ages, Information Services Division, NHS National Services Scotland. June 2008

World Health Organization. Chronic Obstructive Pulmonary Disease (COPD). Avalaible from: http:// www.who.int/respiratory/copd/en/index.html

* Note: There are 14 references in all — Click here to view all references

Outcome

Type	Measure	Description	Time frame	Safety issue
Primary	Number of resources after the implementation of ML (Machine Learning) added to a telemedicine system in readmitted COPD patients (telEPOC).	Number of hospitalizations (hospital base data).; Days of hospital staying ((hospital base data).; Emergency visits (hospital base data).; Readmissions (hospital base data).; Visits to pneumology consultation in last 2 years (hospital base data).	2 years
Primary	Change in quality of life in patients after the implementation of ML (in patients that generate alarms)	-CAT (COPD assessment test): impact of COPD on health status. 8 items (cough, phlegm, chest tightness, breathlessness, limited activities, confidence leaving home, sleeplessness and energy), scaling from 1 to 5. Higher scores denote a more severe impact of COPD on a patient's life.	2 years
Primary	Changes in quality of life in patients after the implementation of ML (in patients that generate alarms)	EuroQol-5d questionnaire: measure of health for clinical and economic appraisal.; 2 parts: 5 dimensions descriptive system (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression). Each of them has 3 levels of severity (no problems -1 point- , some problems -2 points- or moderate-severe problems - 3 points-). Having more points represents a worse situation.; A visual analog scale for a more general evaluation. It is a vertical scale, ranging from 0 (worst imaginable state of health) to 100 (best imaginable state of health). In it, the individual must mark the point on the vertical line that best reflects the assessment of their global health status today.	2 years
Primary	Cost of the implementation of ML in relation to the standard telemonitoring programmes	- Economic evaluation, inlcuding all the interventions carried out inherent to the program, ranging from phone calls, patient displacement for consultation, drug use, hospitalizations and visits to emergencioes, primary and specialized care (hospital base data).	2 years
Primary	Workload of nurses	- The time that must be spend every day managing the alarms after adding ML.	2 years
Primary	Changes in clinical diary practice after including ML	- Exercise capacity (six minutes walking test)	2 years
Primary	Change in clinical diary practice after including ML	- Physical activity (pedometer)	2 years