Lisa M. Vizer

Detecting cognitive and physical stress through typing behavior

Monitoring of cognitive and physical function is central to the care of people experiencing or at risk for various health conditions, but existing solutions rely on intrusive methods that are inadequate for continuous tracking. This research explores the possibility of detecting cognitive and physical stress by monitoring keyboard interactions with the eventual goal of detecting acute or chronic changes in cognitive and physical function. Preliminary results indicate that it is possible to classify cognitive and physical stress conditions relative to non-stress conditions based on keystroke and text features with accuracy rates comparable to those currently obtained using affective computing methods. The proposed approach is attractive because it requires no additional hardware, is unobtrusive, is adaptable to each user, and is very low-cost.

Usability of a Patient Education and Motivation Tool Using Heuristic Evaluation

Background Computer-mediated educational applications can provide a self-paced, interactive environment to deliver educational content to individuals about their health condition. These programs have been used to deliver health-related information about a variety of topics, including breast cancer screening, asthma management, and injury prevention. We have designed the Patient Education and Motivation Tool (PEMT), an interactive computer-based educational program based on behavioral, cognitive, and humanistic learning theories. The tool is designed to educate users and has three key components: screening, learning, and evaluation. Objective The objective of this tutorial is to illustrate a heuristic evaluation using a computer-based patient education program (PEMT) as a case study. The aims were to improve the usability of PEMT through heuristic evaluation of the interface; to report the results of these usability evaluations; to make changes based on the findings of the usability experts; and to describe the benefits and limitations of applying usability evaluations to PEMT. Methods PEMT was evaluated by three usability experts using Nielsen’s usability heuristics while reviewing the interface to produce a list of heuristic violations with severity ratings. The violations were sorted by heuristic and ordered from most to least severe within each heuristic. Results A total of 127 violations were identified with a median severity of 3 (range 0 to 4 with 0 = no problem to 4 = catastrophic problem). Results showed 13 violations for visibility (median severity = 2), 38 violations for match between system and real world (median severity = 2), 6 violations for user control and freedom (median severity = 3), 34 violations for consistency and standards (median severity = 2), 11 violations for error severity (median severity = 3), 1 violation for recognition and control (median severity = 3), 7 violations for flexibility and efficiency (median severity = 2), 9 violations for aesthetic and minimalist design (median severity = 2), 4 violations for help users recognize, diagnose, and recover from errors (median severity = 3), and 4 violations for help and documentation (median severity = 4). Conclusion We describe the heuristic evaluation method employed to assess the usability of PEMT, a method which uncovers heuristic violations in the interface design in a quick and efficient manner. Bringing together usability experts and health professionals to evaluate a computer-mediated patient education program can help to identify problems in a timely manner. This makes this method particularly well suited to the iterative design process when developing other computer-mediated health education programs. Heuristic evaluations provided a means to assess the user interface of PEMT.

Classifying Text-Based Computer Interactions for Health Monitoring

Detecting early trends indicating cognitive decline can allow older adults to better manage their health, but current assessments present barriers precluding the use of such continuous monitoring by consumers. To explore the effects of cognitive status on computer interaction patterns, the authors collected typed text samples from older adults with and without pre-mild cognitive impairment (PreMCI) and constructed statistical models from keystroke and linguistic features for differentiating between the two groups. Using both feature sets, they obtained a 77.1 percent correct classification rate with 70.6 percent sensitivity, 83.3 percent specificity, and a 0.808 area under curve (AUC). These results are in line with current assessments for MC--a more advanced disease--but using an unobtrusive method. This research contributes a combination of features for text and keystroke analysis and enhances understanding of how clinicians or older adults themselves might monitor for PreMCI through patterns in typed text. It has implications for embedded systems that can enable healthcare providers and consumers to proactively and continuously monitor changes in cognitive function.

Different strokes for different folks: individual stress response as manifested in typed text

Stress is a part of everyday life, but chronic high stress can have psychological and physiological side effects. Systems that can detect harmful levels of stress could assist users in managing their stress and health. However, current assessments are often obtrusive or require specialized equipment. This research leverages attributes of everyday keyboard interactions to proactively and continuously monitor cognitive function. A laboratory study was conducted where typing samples were collected under stress and no-stress conditions. Keystroke and linguistic features were extracted from the samples and models were constructed for each participant. Correct classification rates ranged from 62% to 88% with a mean of 72%.

Detecting cognitive impairment using keystroke and linguistic features of typed text: toward an adaptive method for continuous monitoring of cognitive status

Perception, attention, and memory form the foundation of human cognition, and are functions that most people take for granted. However, factors such as environment, mood, stress, education, trauma, aging, or disease can impact cognitive function both positively and negatively. For example, working memory capacity generally declines somewhat with age, but a particular individuals accumulated knowledge and skills usually remain intact and can continue to grow. Current methods of monitoring persons for cognitive decline use only normative data and do not take individual differences into account. Given that early intervention can lessen the impact of cognitive decline, concern that current cognitive assessments do not adequately address individual differences, and growing technology use by older adults, this paper investigates a more effective method for monitoring cognitive function using everyday interactions with IT.

Generations in the Workplace: An Exploratory Study with Administrative Assistants

To better support older adults in the workplace, this study examines the strategies workers employ to learn software and complete tasks. The purpose of the overall research project is to understand how to help older workers adapt to and remain productive in the workplace. This knowledge may inform the design and development of training modules and software extensions to accommodate the needs of workers as they age. This paper describes an exploratory study in which administrative assistants at an industrial research facility were interviewed and surveyed about their work practices, preferences, and attitudes. The data revealed a high level of communication, knowledge sharing, and collaboration among the assistants. Possibilities for future research are inclusion of workers at other companies and in other jobs, examination of the motivations and attitudes surrounding work behavior, and development of design guidelines for software tools.

Efficacy of personalized models in discriminating high cognitive demand conditions using text-based interactions

Although high cognitive demand conditions can impair psychological, physical, and behavioral processes without appropriate management, current measurement methods are too cumbersome for continuous monitoring of cognitive demand, and do not account for individual differences. This research uses keystroke and linguistic markers of typed text to construct individualized models of cognitive demand response to discriminate high and low cognitive demand conditions, the results of which can have implications for design of cognitive demand monitoring systems for personalized health management. We constructed within-subject models of cognitive demand response for nine participants and one between-subjects model based on 20 participants. The AUCs for personalized models ranged from 0.679 to 0.953 (Mean=0.826, SD=0.085), significantly higher than chance (p<0.0001) and the 0.714 AUC for the generic model (p=0.002). Although the features in each model were different, the most common features across models are rate of negative emotion, lexical diversity, rate of words over six letters, and word count. These results confirm significant individual differences in cognitive demand response and suggest that those developing measurement methods used in a monitoring system should consider adaptation to individual characteristics. Our research operationalizes the effects of cognitive demand on HCI and contributes a unique combination of text and keystroke features used to detect high cognitive demand situations. An unobtrusive approach is proposed for classifying high cognitive demand conditions using typing and language features.Individualized models of high cognitive demand response are significantly more accurate than a generic model.Individualized models show significant interpersonal differences in the response to high cognitive demand.

Understanding Design Tradeoffs for Health Technologies: A Mixed-Methods Approach

We introduce a mixed-methods approach for determining how people weigh tradeoffs in values related to health and technologies for health self-management. Our approach combines interviews with Q-methodology, a method from psychology uniquely suited to quantifying opinions. We derive the framework for structured data collection and analysis for the Q-methodology from theories of self-management of chronic illness and technology adoption. To illustrate the power of this new approach, we used it in a field study of nine older adults with type 2 diabetes, and nine mothers of children with asthma. Our mixed-methods approach provides three key advantages for health design science in HCI: (1) it provides a structured health sciences theoretical framework to guide data collection and analysis; (2) it enhances the coding of unstructured data with statistical patterns of polarizing and consensus views; and (3) it empowers participants to actively weigh competing values that are most personally significant to them.

The Patient-Centered Electronic Health Record and Patient Portals

Personal health records (PHR), particularly the patient portal, are touted as instrumental to the future of medical care and health management. The patient portal could be a powerful means to engage patients and empower them to manage their own health. Studies show that patients are interested in patient portals, specifically in the self-management and administrative aspects. However, some studies also show little evidence to support claims of improvement in health outcomes, cost measures, or health care utilization. This chapter presents an overview of personal health records and patient portals, beginning with the evolution of these technologies. We next cover the data and functionality components of personal health records and patient portals. Then, we discuss the context in which personal health records and patient portals are situated, including political forces, Meaningful Use, user access and usability, and data issues. Finally, throughout the chapter we consider future directions for personal health records and patient portals.