Advanced Computational Intelligence Paradigms in Healthcare 5

This chapter presents an overview of some of the most recent advances in the application of decision support systems in healthcare. A summary of the chapters on clinical decision support systems, rehabilitation decision support systems, and some current factors involving technology acceptance is presented.

During recent years, an increasing number of health records have become virtual or electronic. This chapter discusses some of my early work in virtualization of health records in the context of the FDA drug approval process. The chapter also outlines the lesser-known area of government regulation of virtualized or electronic health records governed by HIPAA and related standards. The constraints imposed by these regulations do not impose impossible obstacles, but they do provide constraints that are as real as any other underlying system or design constraint. This chapter then continues to discuss other current systems, such as the Kaiser and veterans health association virtual health records systems, and forward-looking systems, such as Google Health, and how these systems may relate to and be constrained by the current government data standards.

In this paper we make an extensive study of Artificial Intelligence (AI) techniques that can be used in decision support systems in healthcare. In particular, we propose variants of ensemble methods (i.e., Rotation Forest and Input Decimated Ensembles) that are based on perturbing features, and we make a wide comparison among the ensemble approaches. We illustrate the power of these techniques by applying our approaches to different healthcare problems. Included in this chapter is extensive background material on the single classifier systems, ensemble methods, and feature transforms used in the experimental section.

Fast, reliable, and correct medical diagnostics is of utter importance in today’s world where diseases can spread quickly. For this reason, we have developed a medical diagnosis system that is based on multi agent system theory, the holonic paradigm, and swarm intelligence techniques. More specifically, a huge number of comparatively simple agents form the basis of our system. In order to provide a solid medical diagnosis always a set of relevant agents needs to work together. These agents will provide a huge set of possible solutions, which need to be evaluated in order to conclude. The paradigm of swarm intelligence implies that a set of comparatively simple entities produces sophisticated and highly reliable results. In our scenario, it means that our agents are not provided with a real world model; i.e., it has only a very limited understanding on health issues and the process of medical diagnosis. This puts a huge burden on the decision process.

This paper concentrate on the decision process within our system and will present our ideas, which are based on decision theory, and here, especially, on Bayesian probability since, among others, uncertainty is inherent feature of a medical diagnosis process. The presented approach focuses on reaching the optimal medical diagnosis with the minimum risk under the given uncertainty. Additional factors that play an important role are the required time for the decision process and the produced costs.

One of the most relevant aspects in human assisted reproduction is to decide if, at a given moment, the endometrium is receptive for embryo implantation, in order to perform embryo transfer cycle or to postpone it in another cycle. This might increase both patients’ convenience and the cost-effectiveness of the assisted reproduction procedure. To help human experts in taking this decision, we developed an artificial intelligence system based on a data mining approach where data extracted from the endometrium/subendomentrium and their vascularization are evaluated. The proposed system has been tested on a dataset of 62 cycles of intracytoplasmic sperm injection (ICSI) and several machine learning methods are compared for obtaining a high performing system. Particularly interesting is the performance obtained considering only three features: the patient’s age, the subendometrial volume and the endometrial vascularization/flow index; the best system, based on a random subspace of decision tree, obtains an area under the ROC curve (AUC) of 0.85 in predicting the pregnancy rate. These preliminary results show that it is possible to measure in a non invasive way a set of features from a patient, for assisting the decision of making or postponing the embryo transfer.

In recent years, there has been an increasing trend towards using video games for health applications. In particular interactive video games, where an individual interacts with the game by moving their limbs or whole body, have started to find application in the field of rehabilitation medicine. The often dull and repetitive nature of rehabilitation exercise can be transformed into an activity to which patients happily adhere via the use of engaging video games that are enjoyable to play. One additional potential benefit of video game use in rehabilitation is that patients can continue to interact with the video game system in their own home following discharge from hospital. As such, video games may offer a means for rehabilitation specialists to remotely assess compliance of patients with their rehabilitation therapy and monitor changes in function over time. Although the use of technology for monitoring health at home is now widespread, an as yet unexplored challenge lies in integrating information technologies with rehabilitation games. This keeps the health professional informed about compliance and progress of the video game exercise, while the patient performs her/his prescribed rehabilitation routine at home. Therefore, there is a strong need for a computational framework to support the medical professional and patient by using an agent-based architecture. Agents are pieces of software that act on behalf of human roles, involved in rehabilitation process. The objective of this chapter is to thus address major issues in designing an agent-based mobile monitoring system for rehabilitation treatments. The chapter also suggests how to remotely measure the patient’s progress in rehabilitation treatments while the patient plays video games at home.

Intelligent Decision-Support (IDS) mechanisms to improve an ‘in-action’ facilitator intervention model and ‘on-action’ evaluation and refinement model are proposed for contemporary Virtual Reality Healthcare & Rehabilitation training. The ‘Zone of Optimized Motivation’ (ZOOM) model and the ‘Hermeneutic Action Research Recursive Reflection’ model have emerged from a body of virtual reality research called SoundScapes. The work targets all ages and all abilities through gesture-control of responsive multimedia within Virtual Interactive Space (VIS). VIS is an interactive information environment at the core of an open-ended custom system where unencumbered residual function manipulates selected audiovisual and robotic feedback that results in afferent-efferent neural feedback loop closure. Such loop closure is hypothesized as the reason why such interactive system environments are so effective in the context of rehabilitation and healthcare. The approach is adaptive across the range of dysfunction, from the most profoundly disabled to traditionally developed. This proposal considers enhancing VIS data exchange, i.e. human input information matched to responsive content, through dynamic decision-support of adjustment of difficulty encountered. To date facilitator role has included manual parameter manipulation of interface to affect an invisible active zone quality (typically, sensitivity or location) and/or content quality. Inaction human adjustment-decisions are according to interpretation of user state and engagement. Questioned is whether automated support for such decisions is feasible so that dynamic difficulty adjustment (DDA) of that which is encountered by the user is considered optimal to goal. Core issues are presented to detail and justify the concept. Findings are related to current trends with conclusions reflecting on potential impact.

The past few decades have seen incredible development of technology and systems for image-assisted surgery. Adoption of image-assisted surgical systems by medical practitioners, however, has lagged considerably behind the advances in technology. Whereas this lag in adoption is not an unknown phenomenon when it comes to technological advancement, it is more troubling in the context of patient safety and patient health. With the use of certain image-assisted surgical systems, operative time can be greatly reduced, patients outcomes can be improved, and morbidity and mortality rates can be reduced. The question is then: why aren’t the systems being adopted? This chapter discusses a number of issues related to the adoption of surgical trainers and image assisted surgery, including business and reimbursement issues, training issues, and health care liability structures. Examples are taken form the author’s own unadopted advances as well as adopted and unadopted systems from other research and business teams.

This chapter addresses the role of presence or “the perceptual illusion of nonmediation” [26] in user responses to technologies for healthcare, with particular attention to intelligent decision support systems (IDSS). It begins by defining presence and reviewing relevant research on presence-related responses to agents, virtual reality systems, and other technologies. It then discusses the implications of presence to researchers and health practitioners with an interest in IDSS, including a series of recommendations.

Witnessing is core to the design of social interaction. This chapter explores the role of witnessing from different perspectives. The first perspective focuses on witnessing in its social and psychological consequences. Response-ability, address- ability, the performance of testimony and transparency of subject position determine how individuals perceive/ witness each other. The second perspective focuses on the impact of technology on witnessing and introduces the YUTPA framework as a tool for the design and orchestration of witnessing in technology environments. The third, fourth and fifth sections discuss initial results of exploratory research performed in the Netherlands and in India. This research shows that the way in which witnessing is orchestrated affects the psychological wellbeing of the people involved: it can be beneficial or detrimental. These results demonstrate the need to explicitly design witnessing along the four dimensions of the YUTPA model: space, time, action and relation. The sixth section addresses a third perspective, the technological perspective that focuses on the design of large scale socio-technological systems. The conclusion of this chapter argues that health systems that affect the psychological well being of the people involved must both be designed to take witnessing into account but also to be used appropriately.