Speech and Automata in Health Care

Amy Neustein, Amy Neustein

• Publisher: De Gruyter
• Year: 2014
• ISBN:
  9781614519607
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Robotics
• Index: Technology & Engineering

• Examines various speech technologies deployed in healthcare service robots to maximize the robot's ability to interpret user input.
• Demonstrates how robot anthropomorphic features and etiquette in behavior promotes user-positive emotions, acceptance of robots, and compliance with robot requests.
• Analyzes how multimodal medical-service robots and other cyber-physical systems can reduce mistakes and mishaps in the operating room.
• Evaluates various input methods for improving acceptance of robots in the older adult population.
• Presents case studies of cognitively and socially engaging robots in the long-term care setting for helping older adults with activities of daily living and in the pediatric setting for helping children with autism spectrum conditions and metabolic disorders.

Speech and Automata in Health Care forges new ground by closely analyzing how three separate disciplines - speech technology, robotics, and medical/surgical/assistive care - intersect with one another, resulting in an innovative way of diagnosing and treating both juvenile and adult illnesses and conditions. This includes the use of speech-enabled robotics to help the elderly population cope with common problems associated with aging caused by the diminution in their sensory, auditory and motor capabilities. By examining the emerging nexus of speech, automata, and health care, the authors demonstrate the exciting potential of automata, both speech-driven and multimodal, to affect the healthcare delivery system so that it better meets the needs of the populations it serves. This book provides both empirical research findings and incisive literature reviews that demonstrate some of the more novel uses of speech-enabled and multimodal automata in the operating room, hospital ward, long-term care facility, and in the home. Studies backed by major universities, research institutes, and by EU-funded collaborative projects are debuted in this volume.

This volume provides a wealth of timely material for industrial engineers, speech scientists, computational linguists, and for signal processing and intelligent systems design experts.

Topics include:

• Spoken Interaction with Healthcare Robots
• Service Robot Feature Effects on Patient Acceptance/Emotional Response
• Designing Embodied and Virtual Agents for the Operating Room
• The Emerging Role of Robotics for Personal Health Management in the Older-Adult Population
• Why Input Methods for Robots that Serve the Older Adult Are Critical for Usability
• Socially and Cognitively Engaging Robots in the Long-Term Care Setting
• Voice-Enabled Assistive Robots for Managing Autism Spectrum Conditions
• ASR and TTS for Voice-Controlled Robot Interactions in Treating Children with Metabolic Disorders

Part I

The evolution and design of service robots in health care: evaluating the role of speech and other modalities in human-robot interaction

dp n="24" folio="" ? dp n="25" folio="" ?

António Te ixeira

1

A critical analysis of speech-based interaction in healthcare robots: making a case for the increased use of speech in medical and assistive robots

Abstract: Healthcare systems around the world face serious challenges related to an aging population and the lack of enough qualified professionals to serve the needs of the elderly. To meet these challenges, health care must place greater emphasis on effective use of technology. Healthcare robots are viewed as a possible answer, and in fact, more and more service robots are expected to enter the healthcare sector in the near future. To improve acceptance of such robots, it is important to focus on how they interact with humans. Research on Human-robot interaction (HRI) clearly indicates that speech is one of the preferred ways of interacting. This chapter is a position paper championing the importance of developing adequate speech-enabled interfaces for medical and assistive robots. To do so, the author surveys the professional literature detailing the design and use of healthcare robots. Exploring some of the principal challenges presented to system designers, the author also shows how some of those challenges might be addressed. A critique of the current systems and future directions are proposed.

1.1 Introduction

Many countries around the world face serious challenges in the delivery of health care, partly due to the increase in the percentage of the population above 65, and the lack of enough qualified professionals to attend to those individuals. It is now generally accepted that the increased use of technical systems in the healthcare sector will be inevitable to meet these challenges. As a result, health care must place greater emphasis on community care and the effective use of technology, such as the quite promising service robots. As a matter of fact, robots have long captured our imagination and are being used increasingly to assist humans in repetitive and physically demanding work or for things that need accuracy and precision. In such tasks, the use of robots is advantageous because they are potentially quick to train, cheap to maintain, easy to refuel and repair and not prone to be bored by repetitive tasks as humans are. In health care, they can help the elderly or chronically ill to remain independent, reducing the need for health professionals and the demand for in home care.

Robots are not a novelty in the healthcare system. For example, the da Vinci Surgical System (Ballantyne & Moll 2003; DiMaio, Hanuschik & Kreaden 2011; Chen & Yu 2012), has conducted more than 20,000 surgeries (McNickle 2012). “A robot controlled through hand gestures and speech commands is a natural alternative that does not affect the normal flow of surgery” (Jacob et al. 2013). Robots like the Aethon TUG (Aethon 2014) act as a distribution system that move through hospital corridors, elevators and departments to make either scheduled or on-demand deliveries. Outside the health institutions, robots are being used to enhance telemedicine and care for those remaining in their homes.

Although at present healthcare robots are more widely used in the clinical environment, they can potentially become popular for personal use as well. Robots have the capability of assisting the humans in their homes or at a healthcare facility. By using input/output devices such as microphones and cameras, such robots can serve a larger area. This solves the problem of having to have multiple devices positioned at various places in the home or healthcare facility. Some of the tasks fulfilled by such robots are the delivery of objects, food, and drugs. They can also serve as a real companion. Researchers predict that “[i]n the next few years, thousands of ‘service robots’ are expected to enter the healthcare sector” (Hay 2012; McNickle 2012). In fact, service robots have the potential to enhance the quality of life for a broad range of users. It is expected that robotics will play a key role in many challenging areas such as the performance of household chores, and in assisting the aging population and those with physical impairments or those who are undergoing rehabilitation therapy (Tapus, Mataric & Scassellati 2007).

To improve acceptance of service robots, it is important to focus on how they interact with humans. Sung, Christensen and Grinter (2009) show the need for efficiency in human-robot interaction (HRI) and indicate that speech is clearly the preferred mode of interaction. In fact, to demonstrate the importance of speech to the user engaged in HRI, one of the participants of Sung and coworkers study actually drew ears on the robot to show the relevance attributed to this modality of interaction.

The speech preference shown by research subjects should not come as a surprise. First, humans are “wired for speech” (Nass & Brave 2007). Second, some characteristics of spoken language are well fitted for robots: namely, it makes possible communication at a certain distance; it does not require the use of hands; it requires no visual attention; it is natural and so, people communicate as they normally do; and it is fast (commonly 150–250 word per minute) (Bernsen 1997).

A number of experiments show that the human brain rarely makes a distinction on some level between speaking to a man or a machine (Nass & Brave 2007). Thus, speech is the most natural and easy interface to deal with computers, not only for people with special needs, but for people in general, as stated in (Nass & Brave 2007, p. 3):

“Ubiquitous computing – access to all information for anyone, anywhere, at any time – relies on speech for those whose eyes or hands are directed to other tasks (such as driving ...) or for those who cannot read or type (such as children, the blind, or the disabled)”.

All in all, in Human-Computer Interaction (HCI) the state-of-the-art can be characterized by machines that adapt to humans. There is no doubt that speech enhances the way robots adapt to humans. One of the major challenges of the 21^st Century will be the development of different types of robotic systems in the health area that can effectively interact with humans. It is very important for these systems to have good human-robot interaction (HRI), allowing them to be easily accepted, usable, and controllable by humans.

The main objective for this chapter is to help meet this challenge of seamless integration of robots in health care by analyzing the current status of robots in health care and to provide some guidelines for the future. This chapter is a position paper (though the author gives much heed to relevant recent works in his review of the state-of-the-art of robots in health care with regard to role played by speech) championing the importance of developing adequate speech-enabled interfaces of medical and assistive robots. The chapter focuses on health care in its broadest sense, by including assisting elderly at home (as part of Ambient Assisted Living – AAL), the visually impaired, or those who have mobility restrictions. Hence, it doesn’t focus exclusively on just one area of service robots, such as speech-driven medical or surgical robots that help in the performance of delicate procedures. The chapter is organized as follows: The section following this introduction provides relevant background material on the different uses of robots in health care, on speech-based interaction with machines and the main technologies enabling such interaction. The section that follows presents a brief review of concrete usage of spoken interaction with healthcare robots. This review is followed by a critical discussion, based on the well-known SWOT method.¹ The chapter concludes with both a discussion of the existing technologies and a presentation of a roadmap for future research.

1.2 Background

To contribute to an easier understanding of the review and critical assessment of the state-of-the-art regarding the use speech in the creation of natural, usable and accessible interaction with healthcare robots, in this section the author provides essential information on the two basic components of robots in health care: First, on healthcare robots and their multiple areas of application; and second on speech-based interaction and the technologies that support it.

1.2.1 Robots and health care

Many different types of robots exist. Figure 1.1 provides a classification of these different kinds of robots. In the broadest sense, there are three distinct types of robots: military, industrial and service robots. Service robots can be divided into personal and professional subtypes, both of which are relevant to health care. As part of professional service robots we have medical and health service robots, which are used for surgery, diagnostic, rehabilitation, and assistance; and from the personal services side we have robots serving as personal assistants, which are used to aid persons with disabilities, seniors with age-related problems, and in eHealth. In this section, concrete applications of both professional and personal robots are highlighted.

Fig. 1.1: Robots classification. Types and sub-types that are relevant for health care are identified by darker circle outlines.

Service robots are being used in varied healthcare tasks, ranging from surgical droids that can suture a wound to “nanobots” that can swim in the bloodstream. Their “job” can be performed in different settings, from hospital operating rooms to patients’ homes. We discuss the two types of service robots below:

The professional service robots, usually found at healthcare institutions, such as hospitals, are used to improve the quality of medical care received by patients, improve patient and staff satisfaction, and reduce costs. Concrete applications are:

• – Logistic and support tasks: Vecna Medical, for example, has developed QC Bot(R) (VECNA 2013) that navigates its way through complex hospital campuses, both indoors and out, to deliver and transport medical supplies, medications, and even meals. It also allows for telemedicine and teleconference functions and some patient self-service functions such as check-ins and bedside registration. Another example is the Aethon TUG (Aethon 2014), an automated system that allows a facility to move supplies such as medication, linens and food from one place to another.
• – Clinical tasks: Robots are being used in several surgical procedures, such as, for example, robot-assisted Thoracoscopic Lymphadenectomy (Suda et al. 2012). They also assist in the work of other health professionals such as nurses (Jacob et al. 2013).
• – Physical Rehabilitation: For example, Toyota announced four robots made to help paralyzed patients walk or balance themselves (McNickle 2012). The robot acts as a two-wheeled balancing game. The machine displays one of three sports games on a monitor and requires the patient to make moves in the game by shifting his/her weight. Other medical robots developed by Toyota include The Walk Training Assist robot (McNickle 2012) and the Independent Walk Assist robot. The robot helps the knee swing and the leg move forward to facilitate walking. Rehabilitation robotics have also been developed, for example, to aid in recovery after a stroke (Wagner et al. 2011).
• – Companions: (Csala, Nemeth & Zainko 2012) presents the application of NAO humanoid robots in a Children’s Hematology and Stem Cell Transplantation Unit where it acts as a companion to cheer children up and break their usual daily routine with performances and exercise.
• – Care: Ranging from robots for psycho-geriatric care of patients with dementia such as PARO (Gelderblom et al. 2010; de Sant’Anna, Morat & Rigaud 2012; Inoue, Wada & Uehara 2012; Chang, Sabanovic & Huber 2013) to robot-assisted play for children with cognitive disabilities (Robins et al. 2012). Most of these robots, if costs decrease, could also integrate the “nonprofessional” personal service robots.

The personal service robots, or assistant robots, address more personal task, such as:

• – Persons with disabilities: A representative example is a guide-dog robot system for visually impaired, providing multiple functions for the self-walking in urban systems, such as following, navigation and obstacle avoidance (Wei, Kou & Lee 2013).
• – Robots for virtual presence, telemedicine and eHealth: A representative example from 2012, developed by robotics firm iRobot in collaboration with InTouch Health, is the Remote Presence Virtual + Independent Telemedicine Assistant, or RP-VITA (McNickle 2012), which combines iRobot’s telepresence units with InTouch health’s distance education tools, creating a system that allows physicians to care for patients remotely. It includes mapping and obstacle detection, an iPad user interface for control and interaction and can interface with diagnostic devices and electronic medical records (EMR) systems.
• – Service robots for aging-in-place: Examples are Care-O-bot, a multifunctional assistant using a graphical user interface and speech, which is operated by the elderly person living independently at home (Schraft, Schaeffer & May 1998; Hans & Baum 2001; Graf et al. 2002; Reiser et al. 2009); Telerobot, a remotely operated robot equipped with video conferencing capability for telerehabilitation at home (Brière, Boissy & Michaud 2009); BIRON (Bielefeld Robot companION) also developed for home use (Haasch et al. 2004). The potential of robots in this area was recently reviewed in Bemelmans et al. (2011). A systematic review of literature was performed to assess the effects of the interaction of elderly with socially assistive robots. The authors found studies reporting positive effects of companion robots, both in terms of psychological and physiological benefits.

1.2.2 Speech-based interaction with machines

By adopting the definition of modality as “a way of exchanging information between humans […] and machines, in some medium” (Bernsen 2002), several speech modalities can be considered. In the Bernsen taxonomy three modalities are proposed at the atomic² level: 1) spoken discourse, 2) spoken label/keywords, and 3) spoken notation (Bernsen & Dybkjaer 2009, 2010).

Spoken discourse is well known and the most used modality for communication among humans. Spoken labels/keywords refer to small units used to convey very limited/isolated pieces of meaning. Since they have a limited grasp and are used outside a linguistic text, the lack of context might sometimes result in ambiguity. Furthermore, ambiguity might also result from how different people, even sharing a common language, use different keywords to address the same objects or situations (e.g., cab/taxi, holiday/vacation, etc.). Spo...