The Future of Applications For Clinical Decision Support Systems in Healthcare. Case Study: H2020 PERSIST Project

Abstract

Clinical decision support systems (CDSS) are computer-based systems that leverage medical knowledge and patient-specific data to respond to a request for decision support by providing recommendations to improve the quality of the service provided to the patient. A complete and successful CDSS has a complex structure and includes various components from different scientific disciplines. Thus developers should take into account all faces of the CDSS while implementing the CDSS. CDSS mainly consists of three components like a rule engine, an inference engine, and a mechanism to communicate. The rule engine directly depends on the data types which are based on a knowledge base, a non-knowledge base (artificial intelligence (AI) algorithms), or a hybrid of them. Although AI algorithms have become very popular in the healthcare sector as others, there are some restrictions and obstacles to use them directly in the health-related data and to gain the trustworthiness of the clinicians. One of the solutions to overcome these difficulties is to use the explainable AI (XAI) methods. An effective CDSS needs to be fed by analysis of clinical or health-related data for constructing a knowledge base. Such data may include sensitive/personal information that needs to be anonymized and to be prevented from re-identification. This means that CDSS should comply with privacy/ethical concerns besides the technical requirements. On the other hand, wearable technology which plays an important role in the functioning of the CDSS is becoming popular as a real-world data collector for heterogeneous and complex health-related data to support more accurate and personalized decisions for both patients and clinicians. Also, people using these smart devices in connection with CDSS via smart apps easily benefits from healthcare technologies like telemonitoring or telehealth. The importance of these kinds of technologies appeared more clearly nowadays. Because the COVID-19 situation showed us that people might have limited access to health services and have difficulties in all stages of treatment. The PERSIST (Patient-centered SurvivorShIp care plan after Cancer treatments based on Big Data and Artificial Intelligence technologies) project is an example approved H2020 project for all mentioned requirements for CDSS. PERSIST project aims at developing an open and interoperable ecosystem to improve the care of cancer survivors. PERSIST will advance the maturity level of existing algorithms that estimate the risk of re-identification in anonymized datasets. The role of CDSS in the PERSIST project is to provide patient-specific and actionable recommendations to health care providers for timely identification of symptoms and/or risks associated with the development of new medical conditions or impairments in the quality of life (QoL) of the patient. Soon, CDSS will have a central role in the healthcare system with the help of algorithmic advances. This paper proposes possible directions of CDSS in healthcare while discussing the advances in the technology and the challenges for developing such systems by giving an example case study.

Full text

IES’20 International Engineering Symposium | Engineering Applications in Industry (Virtual)
December, 5-6 and 10-13, 2020 | Izmir Democracy University, Izmir, Turkey
The Future of Applications For Clinical Decision Support Systems in Healthcare.
Case Study: H2020 PERSIST Project
Umut ARIÖZ
1
, Barış YILDIZ
2
, R. Alp KUT
3
, İbrahim Tolga AĞIM
1
, Kadir ÜĞÜDÜCÜ
1
1 EMODA Yazılım ve Danı
ş
manlık. San ve Tic Ltd
Ş
ti,IYTE Teknopark Urla
İ
zmir,
umut@emodayazilim.com

,

info@emodayazilim.com

, project@emodayazilim.com
2 Department of Software Engineering, Ya
ş
ar Üniversitesi Selçuk Ya
ş
ar Yerle
ş
kesi, Bornova
İ
zmir,
baris.yildiz@yasar.edu.tr
3 Semafor Teknoloji Yazılım Danı
ş
manlık Ltd.
Ş
ti, Dokuz Eylül Üniversitesi
İ
nciraltı Yerle
ş
kesi, Balçova,
İ
zmir, alp.kut@semaforteklonoji.com
ABSTRACT
Clinical decision support systems (CDSS) are computer-based systems that leverage medical
knowledge and patient-specific data to respond to a request for decision support by providing
recommendations to improve the quality of the service provided to the patient. A complete and
successful CDSS has a complex structure and includes various components from different scientific
disciplines. Thus developers should take into account all faces of the CDSS while implementing the
CDSS.
CDSS mainly consists of three components like a rule engine, an inference engine, and a
mechanism to communicate. The rule engine directly depends on the data types which are based on a
knowledge base, a non-knowledge base (artificial intelligence (AI) algorithms), or a hybrid of them.
Although AI algorithms have become very popular in the healthcare sector as others, there are some
restrictions and obstacles to use them directly in the health-related data and to gain the
trustworthiness of the clinicians. One of the solutions to overcome these difficulties is to use the
explainable AI (XAI) methods.
An effective CDSS needs to be fed by analysis of clinical or health-related data for constructing
a knowledge base. Such data may include sensitive/personal information that needs to be anonymized
and to be prevented from re-identification. This means that CDSS should comply with privacy/ethical
concerns besides the technical requirements.
On the other hand, wearable technology which plays an important role in the functioning of
the CDSS is becoming popular as a real-world data collector for heterogeneous and complex
health-related data to support more accurate and personalized decisions for both patients and
clinicians. Also, people using these smart devices in connection with CDSS via smart apps easily
benefits from healthcare technologies like telemonitoring or telehealth. The importance of these kinds
of technologies appeared more clearly nowadays. Because the COVID-19 situation showed us that
people might have limited access to health services and have difficulties in all stages of treatment.
The PERSIST (Patient-centered SurvivorShIp care plan after Cancer treatments based on Big
Data and Artificial Intelligence technologies) project is an example approved H2020 project for all
mentioned requirements for CDSS. PERSIST project aims at developing an open and interoperable
ecosystem to improve the care of cancer survivors. PERSIST will advance the maturity level of existing
algorithms that estimate the risk of re-identification in anonymized datasets. The role of CDSS in the
PERSIST project is to provide patient-specific and actionable recommendations to health care
providers for timely identification of symptoms and/or risks associated with the development of new
medical conditions or impairments in the quality of life (QoL) of the patient.
Soon, CDSS will have a central role in the healthcare system with the help of algorithmic
advances. This paper proposes possible directions of CDSS in healthcare while discussing the advances
in the technology and the challenges for developing such systems by giving an example case study.
Keywords: Clinical decision support systems, knowledge base, wearable technology, explainable AI