Elsevier

Expert Systems with Applications

Volume 39, Issue 12, 15 September 2012, Pages 10533-10543
Expert Systems with Applications

Sentic PROMs: Application of sentic computing to the development of a novel unified framework for measuring health-care quality

https://doi.org/10.1016/j.eswa.2012.02.120 Get rights and content

Abstract

Barriers to use health related quality of life measuring systems include the time needed to complete the forms and the need for staff to be trained to understand the results. An ideal system of health assessment needs to be clinically useful, timely, sensitive to change, culturally sensitive, low burden, low cost, involving for the patient and built into standard procedures. A new generation of short and easy-to-use tools to monitor patient outcomes on a regular basis has been recently proposed. These tools are quick, effective and easy to understand, as they are very structured and rigid. Such structuredness, however, leaves no space to those patients who would like to say something more. Patients, in fact, are usually willing to express their opinions and feelings in free text, rather than simply filling in a questionnaire, for either speaking out their satisfaction or for cathartic complaining. Sentic PROMs allow patients to evaluate their health status and experience in a semi-structured way and accordingly aggregate input data by means of sentic computing, while tracking patients’ physio-emotional sensitivity.

Highlights

► Develop a novel unified framework for measuring health-care quality. ► Bridge the gap between structuredness and unstructuredness of natural language data. ► Evaluate patients’ health status and experience in a semi-structured way.

Introduction

Public health measures such as better nutrition, greater access to medical care, improved sanitation and more widespread immunization, have produced a rapid decline in death rates in all age groups. Since there is no corresponding decline in birth rates, however, the average age of population is increasing exponentially. If we want health services to keep up with such monotonic growth, we need to automatize as much as possible the way patients access the health-care system, in order to improve both its service quality and timeliness. Everything we do that does not provide benefit to patients or their families, in fact, is waste.

To this end, a new generation of short and easy-to-use tools to monitor patient outcomes and experience on a regular basis has been recently proposed by Benson et al. (2010). Such tools are quick, effective and easy to understand, as they are very structured. However, they leave no space to those patients who would like to say something more.

Patients, in fact, are usually keen on expressing their opinions and feelings in free text, especially if driven by particularly positive or negative emotions. They are often happy to share their health-care experiences for different reasons, e.g., because they seek for a sense of togetherness in adversity, because they benefited from others’ opinions and want to give back to the community, for cathartic complaining, for supporting a service they really like, because it is a way to express themselves, because they think their opinions are important for others. When people have a strong feeling about a specific service they tried, they feel like speaking it out. If they loved it, they want others to enjoy it. If they hated it, they want to warn others away.

Standard patient reported outcome measures (PROMs) allow patients to easily and efficiently measure their health related quality of life (HRQoL) but, at the same time, they limit patients’ capability and will to express their opinions about particular aspects of the health-care service that could be improved or important facets of their current health status. The framework developed within this work, in turn, exploits the ensemble application of standard PROMs and sentic computing (Cambria & Hussain, 2012), a novel approach to opinion mining and sentiment analysis, to allow patients to evaluate their health status and experience in a semi-structured way, i.e., both through a fixed questionnaire and through free text.

The structure of the paper is as follows: Section 2 provides some background about HRQoL measurement, Section 3 explains in detail sentic computing tools and techniques adopted in this work, Section 4 illustrates the processes for the extraction of cognitive and affective information from patient opinions, Section 5 shows how such information can be exploited for monitoring patients’ physio-emotional sensitivity, Section 6 presents a preliminary evaluation of the system and Section 7 comprises concluding remarks and a description of future work.

Section snippets

Related work

In health-care, it has long been recognized that, although the health professional is the expert in diagnosing, offering help and giving support in managing a clinical condition, the patient is the expert in living with that condition. Next-generation patients are central to understanding the effectiveness and efficiency of services and how they can be improved. PROMs provide a means of gaining an insight into the way patients perceive their health and the impact that treatments or adjustments

Sentic computing

Existing approaches to automatic identification and extraction of opinions and sentiments from text can be grouped into three main categories: keyword spotting (Elliott, 1992, Ortony et al., 1988, Wiebe et al., 2005), in which text is classified into categories based on the presence of fairly unambiguous affect words, lexical affinity (Rao and Ravichandran, 2009, Somasundaran et al., 2008, Stevenson et al., 2007, Wilson et al., 2005), which assigns arbitrary words a probabilistic affinity for a

Structuring the unstructured

Among the benefits of questionnaires’ structuredness, there are the quickness, effectiveness and ease to use and understand. However, such structuredness involves some drawbacks. A questionnaire, in fact, can limit the possibility to discover new important patterns in the input data and can constrain users to omit important opinions that might be valuable for measuring service quality. In the medical sphere, in particular, patients driven by very positive or very negative emotions are usually

Monitoring patients’ physio-emotional sensitivity

The importance of physio-emotional sensitivity in humans has been proven by recent health research, which has shown that individuals who feel loved and supported by friends and family, or even by a loving pet, tend to have higher survival rates following heart attacks than other cardiac patients who experience a sense of social isolation. Such concept is also reflected in natural language as we use terms such as ‘heartsick’, ‘broken-hearted’ and ‘heartache’ to describe extreme sadness and

Evaluation

Since Sentic PROMs interface is still under-development, at the time of writing this article we had no consistent dataset of aggregated patient data available for thoroughly testing the system. Hence, a preliminary evaluation of the system had to be performed separately at two different levels: structured level (questionnaire data) and unstructured level (natural language data). As for the structured-level evaluation, a validation study was undertaken to examine the psychometric properties and

Conclusion and future work

Medicine is finally waking up to the use of novel technologies to listen to the ‘wisdom of the patient’. Health-care of the future will be based on community, collaboration, self-caring, co-creation and co-production using technologies delivered via the Web. Engaging patients in their health-care and encouraging people to take responsibility for protecting their health, in fact, are seen as the best way to ensure the sustainability of health systems (WHO, 2000). Patients can play a distinct

Acknowledgments

This work has been part-funded by the UK Engineering and Physical Sciences Research Council (EPSRC Grant Reference: EP/G501750/1), the UK Technology Strategy Board (TSB Grant Reference: 12074-75246) and Sitekit Solutions Ltd. (a Company Registered in Scotland: SC116007).

References (52)

  • R. Brooks

    EuroQoL – the current state of play

    Health Policy

    (1996)
  • H. Park et al.

    A simple and fast algorithm for k-medoids clustering

    Expert Systems with Applications

    (2009)
  • A. Abbasi et al.

    Sentiment analysis in multiple languages: Feature selection for opinion classification in web forums

    ACM Transactions on Information Systems

    (2008)
  • T. Benson et al.

    Evaluation of a new short generic measure of health status

    Informatics in Primary Care

    (2010)
  • M. Bergner et al.

    The sickness impact profile: Conceptual formulation and methodology for the development of a health status measure

    International Journal of Health Services

    (1976)
  • Cambria, E., Hussain, A., Havasi, C., & Eckl, C. (2009). AffectiveSpace: Blending common sense and affective knowledge...
  • Cambria, E., Hussain, A., Durrani, T., Havasi, C., Eckl, C., & Munro, J. (2010). Sentic computing for patient centered...
  • Cambria, E., Song, Y., Wang, H., & Hussain, A. (2011). Isanette: A common and common sense knowledge base for opinion...
  • Cambria, E., Havasi, C., & Hussain, A. (2012). SenticNet 2: A semantic and affective resource for opinion mining and...
  • Cambria, E., Grassi, M., Hussain, A., & Havasi, C. (2011). Sentic computing for social media marketing, Multimedia...
  • E. Cambria et al.

    Sentic computing: Techniques, tools, and applications

    (2012)
  • E. Cambria et al.

    Common sense computing: From the society of mind to digital intuition and beyond

  • E. Cambria et al.

    Sentic medoids: Organizing affective common sense knowledge in a multi-dimensional vector space

  • A. Coulter

    Engaging patients in healthcare

    (2011)
  • Coulter, A., & Ellins, J. (2006). The quality enhancing interventions project: Patient-focused interventions. The...
  • A. Culyer et al.

    Social indicators: Health

    Social Trends

    (1971)
  • A. Donabedian

    Evaluating the quality of medical care

    The Millbank Memorial Fund Quarterly

    (1966)
  • R. Duda et al.

    Pattern classification and scene analysis

    (1973)
  • C. Eckart et al.

    The approximation of one matrix by another of lower rank

    Psychometrika

    (1936)
  • Elliott, C. D. (1992). The affective reasoner: A process model of emotions in a multi-agent system, Ph.D. thesis....
  • S. Fanshel et al.

    A health status index and its application to health-services outcomes

    Operations Research

    (1970)
  • Goertzel, B., Silverman, K., Hartley, C., Bugaj, S., & Ross, M. (2000). The Baby Webmind project. In Proceedings of...
  • J. Hartigan et al.

    Algorithm AS 136: A k-means clustering algorithm

    Journal of the Royal Statistical Society

    (1979)
  • Havasi, C., Speer, R., & Alonso, J. (2007). ConceptNet 3: A flexible, multilingual semantic network for common sense...
  • Havasi, C., Speer, R., Holmgren, J. (2010). Automated color selection using semantic knowledge. In Proceedings of AAAI...
  • Havasi, C., Speer, R., & Pustejovsky, J. (2010). Coarse word-sense disambiguation using common sense. In Proceedings of...
  • Cited by (0)

    View full text