Carlos Fernández-Llatas, Ph.D.

Short Bio

Dr. Carlos Fernández-Llatas is the Head of the Process Mining 4 Health Lab and Deputy Director at SABIEN Group at ITACA institute at Universitat Politècnica de València (Spain), Affiliated Researcher at Karolinska Instituted (Sweden) and Permanent Researcher at the joint research unit in ICT applied to Reengineering socio-sanitary process at Hospital La Fe of Valencia. He received the Ph.D. degree in Computer Science in the Pattern Recognition and Artificial Intelligence Program of that university. He is member of the IEEE CIS Task Force on Process Mining. He participated in more than 30 projects through IV, V VI and VII European Framework program, H2020 program and Spanish Government funded projects. He has published more than 100 scientific papers. He has been member of the Organizing Committee in more than 10 international conferences and member of the Scientific Committee in more than 30. He is reviewer in several indexed journals in Bioengineering and Medical Informatics. His research is mainly focused in the use and promotion of Process Mining technologies as well as Process Management, representation and execution techniques for their application in health and human behavior modelling. The author has applied Interactive Process Mining methodology in more the 18 real medical cases in more than 10 centers over the world and is currently working in 4 EU projects applying Process Mining in Healthcare Domain. In these medical cases, we have analyzed a total more than 1 Billion of log events of more than 25 Millions of patients.

Speech: Towards the application of Interactive Process Mining is actual medicine.

Data Science technologies are regarded as having a key role in the extraction of information from existing databases, personal devices and other information sources to support health professionals in their daily decisions  The analysis of the large amount of data available in hospitals that can be used to support the optimization of clinical processes is an ongoing challenge. These technologies enable health professionals in the creation of better care processes for the improvement of the Quality of Care provided to patients, the effectiveness of the treatments and their cost-effectiveness to patients and such improvements will enhance the sustainability of the health system. Improvements in clinical process management not only save lives but also support the provision of better and personalized care to more patients. However, typical Data Science solutions are shown as black boxes by professionals. This affects the confidence of professionals who are suspicious of the results offered by these technologies. In addition, these inductive methods are based on statistical frameworks that produce accurate results only when the number of cases is adequate. However, clinicians are unlikely to need more support in the standard case because is usually covered by standard treatment. Clinicians require help with treating rare cases and classic Data Science technologies have poorer accuracy in these contexts. In addition to that, Machine Learning techniques are not error free. This error is produced by statistical analysis, so is greater in not common cases. This suppose that usually this techniques has a good precision in common cases and a bad prediction on rare cases. For Health professionals, the most interesting cases are the rare cases, because are the one that produce doubts in the expert because the common case, where the system is precise, is usually covered by the standard treatment.  That means that when the case is more interesting, the system produce a higher error.

In this talk, we will present in the use of  Interactive Process Mining technologies and practical examples of its use in real cases. This technique have been used for discover and analyses clinical pathways associated to emergency department circuits in a human understandable and exploratory way. This technique can highlight the special and essential  characteristics of the process, supporting the experts in the actual understanding of the process. Process Mining techniques are focused on presenting models in an expert understandable way. This, on one hand, solve the black box problems, making experts conscientious of the effects of the decision taken, and, on the other hand, allows adapt the solutions to the special situation supporting in real time the decision of the expert.