Skip to main content

Automated analysis of qualitative data using AI for patient safety


Key facts

Type of research degree
Application deadline
Monday 10 June 2024
Project start date
Tuesday 1 October 2024
Country eligibility
UK only
Source of funding
External organisation
Dr Jonathan Benn
Additional supervisors
Prof Alex Gillespie, Prof Carl Macrae, Dr Luke Budworth
School of Psychology
<h2 class="heading hide-accessible">Summary</h2>

Do you want to apply data science methods to analyse healthcare staff and patient narratives concerning experiences in healthcare systems? &nbsp;Do you want to pioneer innovative new approaches to understanding quality and safety in healthcare? &nbsp;Do you want to develop AI-based patient safety intelligence solutions to support policy-makers and providers as they shape future healthcare systems? &nbsp;If so, this studentship opportunity will appeal to you.<br /> <br /> The NIHR Yorkshire and Humber Patient Safety Research Collaborative (PSRC) is inviting applications for a full-time PhD studentship, to start in 2024, as part of its interdisciplinary Safety Intelligence work stream.&nbsp;This research studentship seeks to bring together data science and safety science, with the aim of enhancing patient safety intelligence in healthcare, based upon automated mining of large qualitative datasets generated by staff and patients. The studentship will appeal to candidates from a range of backgrounds with specific interest in application of AI/machine learning methods for automated language analysis of free text (e.g. sentiment analysis; natural language processing using large &amp; small language models, topic modelling, and similar). It may additionally appeal to candidates with a background in qualitative methods who are interested in extending qualitative approaches using quantitative methods.

<h2 class="heading hide-accessible">Full description</h2>

<p>Patient safety is an important field of health services research that focuses on promoting safety by reducing adverse events and potentially harmful variations in care. Contemporary safety science perspectives additionally aim to build system resilience through learning from highly reliable systems and how care systems and professionals adapt in response to dynamic risks and anticipated threats to safety.</p> <p>Two significant challenges exist for safety intelligence in this area. Firstly, the problem of detecting novel or emerging risks in unstructured data (for example non-safety specific data such as patient experience feedback), and secondly, how to develop actionable safety intelligence from large volumes of free-text data (including safety-specific datasets such as those generated by reporting and surveillance systems; examples include patient safety event reporting and haemovigilance). A limitation of current safety reporting and surveillance schemes is that the rapidly-growing free text datasets they generate, which often contain important signals for learning about how to improve patient safety, typically require considerable investment of resources to analyse, in the form of human-intensive coding with input of subject-matter expertise (Fong, 2021).</p> <p>In patient experience feedback and non-safety specific data sets, natural language analysis may be useful in detecting novel patient safety risks. In work by Gillespie and Reader (2022), automated language analysis of online patient experience feedback was established to be effective in identifying undetected patient safety events. The automated model developed through this work had good precision and excellent recall in identifying patient safety incidents, particularly those that were reported to be unnoticed or unresolved. Furthermore, identified incidents were largely independent of staff-reported incidents and predicted hospital mortality rates. Similarly, natural language analysis of electronic health records has demonstrated good sensitivity for detection of postoperative complications, compared with patient safety indicators (Murff, 2011). In medication safety, NLP might automate the identification of adverse drug events in large-scale adverse event programmes and diverse data sources (Wong, 2018).</p> <p>Approaches to automate analysis may be either framework- or data-driven. Topic modelling of patient concerns data, a data-driven approach, has been shown to perform well compared with human categorisation, enhancing the potential for insight and action (Fairie, 2021). Regarding safety-specific data sources such as adverse event reports, NLP and related approaches may play an important role in enhancing learning through automated classification of event characteristics (Young, 2019). Beyond event-driven narratives, safety reporting systems that collect data on near misses, adaptations, emerging risks and safety improvement suggestions are an additional source of safety wisdom, which might be coded to generate insights into system resilience. Within the broader field of safety science, a wealth of safety evaluation frameworks and contributory factor taxonomies exist and there is considerable scope to develop NLP approaches to automate classification in safety intelligence applications (Fong, 2021; Tabaie, 2023).&nbsp;</p> <p>The research programme and thesis may focus on any combination of related aims, examples including:</p> <ul> <li>Establish the feasibility, utility and potential use cases for future safety intelligence solutions based upon technologies for automated analysis of qualitative data</li> <li>Review and develop classificatory frameworks and taxonomies to guide analysis of free text narrative data based upon contemporary safety science theory and models</li> <li>Develop and test machine learning algorithms to enhance the capacity to develop actionable safety intelligence from qualitative datasets</li> <li>Explore the opportunities to derive signals for learning from diverse sources of data beyond event-driven narratives</li> <li>Compare data-driven vs framework-driven automated topic classification for free-text analysis of patient safety narratives</li> </ul> <p>The studentship will be based across the University of Leeds (with links to the Leeds Institute for Data Analytics) and the Patient Safety Research Collaborative (PSRC) located at the Bradford Institute for Health Research. The student will join a multidisciplinary team of researchers in Patient Safety Intelligence with expertise in data science, health and social science research methods. The supervision team includes expertise in patient safety research (specifically safety intelligence), safety science and artificial intelligence in health care (including application of NLP methods in safety-related contexts). The team comprises Dr Jonathan Benn (School of Psychology, University of Leeds, and theme lead for Safety Intelligence, Yorkshire and Humber PSRC), Prof. Alex Gillespie (Department of Psychological and Behavioural Science, London School of Economics), Prof. Carl Macrae (Nottingham University Business School) and Dr Luke Budworth (Senior Research Data Analyst, Yorkshire and Humber PSRC).</p> <p>Yorkshire and Humber PSRC PhD students will become NIHR trainees and you will benefit from a range of training support and resources to develop your knowledge and health research skills. You will also be embedded within the Yorkshire Quality and Safety Research Group, which is a friendly and dynamic group of researchers conducting high-quality, rigorous and applied health services research. YH PSRC is a collaboration between the Bradford Teaching Hospitals Foundation Trust and the Universities of Leeds and Bradford. Our mission is to deliver research to make healthcare safer. We are one of six NIHR Patient Safety Research Collaborations in England. Our work draws on the knowledge and expertise of well-established networks of researchers, patients, carers, clinicians and industry partners to develop ideas that address patient safety problems. Our research focusses on four themes: Safer systems, culture and practice; De-cluttering (safely) for safety; Supporting safe care in the home; and Rethinking safety intelligence for improvement.</p> <h5>References:</h5> <ul> <li>Fairie P, Zhang Z, D&#39;Souza AG, Walsh T, Quan H, Santana MJ. Categorising patient concerns using natural language processing techniques. BMJ Health Care Inform. 2021 Jun;28(1):e100274. doi: 10.1136/bmjhci-2020-100274. PMID: 34193519; PMCID: PMC8246286.</li> <li>Fong, Allan MS. Realizing the Power of Text Mining and Natural Language Processing for Analyzing Patient Safety Event Narratives: The Challenges and Path Forward. Journal of Patient Safety 17(8):p e834-e836, December 2021. | DOI: 10.1097/PTS.0000000000000837</li> <li>Gillespie A, Reader TW. Online patient feedback as a safety valve: An automated language analysis of unnoticed and unresolved safety incidents. Risk Anal. 2022 Aug 9. doi: 10.1111/risa.14002. Epub ahead of print. PMID: 35945156.</li> <li>Murff HJ, FitzHenry F, Matheny ME, Gentry N, Kotter KL, Crimin K, Dittus RS, Rosen AK, Elkin PL, Brown SH, Speroff T. Automated identification of postoperative complications within an electronic medical record using natural language processing. JAMA. 2011 Aug 24;306(8):848-55. doi: 10.1001/jama.2011.1204. PMID: 21862746.</li> <li>Tabaie A, Sengupta S, Pruitt ZM, Fong A. A natural language processing approach to categorise contributing factors from patient safety event reports. BMJ Health Care Inform. 2023 May;30(1):e100731. doi: 10.1136/bmjhci-2022-100731. PMID: 37257922; PMCID: PMC10254979.</li> <li>Wong A, Plasek JM, Montecalvo SP, Zhou L. Natural Language Processing and Its Implications for the Future of Medication Safety: A Narrative Review of Recent Advances and Challenges. Pharmacotherapy. 2018 Aug;38(8):822-841. doi: 10.1002/phar.2151. Epub 2018 Jul 22. PMID: 29884988.</li> <li>Young IJB, Luz S, Lone N. A systematic review of natural language processing for classification tasks in the field of incident reporting and adverse event analysis. Int J Med Inform. 2019 Dec;132:103971. doi: 10.1016/j.ijmedinf.2019.103971. Epub 2019 Oct 5. PMID: 31630063.</li> </ul>

<h2 class="heading">How to apply</h2>

<p>To apply for this scholarship opportunity applicants should complete an <a href="">online application form</a> and attach the following documentation to support their application.&nbsp;</p> <ul> <li>A full academic CV</li> <li>Degree certificate and transcripts of marks</li> <li>Evidence that you meet the University&#39;s minimum English language requirements (if applicable)</li> </ul> <p>To help us identify that you are applying for this scholarship project please ensure you provide the following information on your application form;</p> <ul> <li>Select PhD in Psychology as your programme of study</li> <li>Give the full project title and name the supervisors listed in this advert</li> <li>For source of funding please state you are applying for a&nbsp;NIHR PSRC&nbsp;Scholarship</li> </ul> <p>If English is not your first language, you must provide evidence that you meet the University&#39;s minimum English language requirements (below).</p> <p><em>As an international research-intensive university, we welcome students from all walks of life and from across the world. We foster an inclusive environment where all can flourish and prosper, and we are proud of our strong commitment to student education. Across all Faculties we are dedicated to diversifying our community and we welcome the unique contributions that individuals can bring, and particularly encourage applications from, but not limited to Black, Asian, people who belong to a minority ethnic community, people who identify as LGBT+ and people with disabilities. Applicants will always be selected based on merit and ability.</em></p>

<h2 class="heading heading--sm">Entry requirements</h2>

Applicants to this scholarship should normally have (or be about to obtain) an Undergraduate degree of 2:1 or above (or international equivalent) in a relevant subject area. A Master&rsquo;s degree is desirable, but not essential (for example: a Master&rsquo;s degree in computer/data science, or a health/medicine-related area such as epidemiology, public health, psychology or medical research). Additionally, experience with programming and in relevant areas of data science, such as machine learning and language models, would be considered an advantage. Strong verbal and written communication skills are required for effective interdisciplinary collaboration and engagement with a broad range of research stakeholders including patients and the public. Applicants who are uncertain about the requirements for a particular research degree are advised to contact the School or Admissions Team prior to making an application.<br /> <br /> Other conditions:<br /> &bull; Due to limited funding, we can only consider applicants for this position who are eligible for UK fee status.<br /> &bull; Applicants must not have already been awarded or be currently studying for a doctoral degree.<br /> &bull; Awards must be taken up by 1st October 2024.<br /> &bull; Applicants must live within a reasonable distance of the University of Leeds whilst in receipt of this scholarship.

<h2 class="heading heading--sm">English language requirements</h2>

Candidates whose first language is not English must provide evidence that their English language is sufficient to meet the specific demands of their study. The minimum English language entry requirement for postgraduate research study in the School of Medicine is an IELTS of 6.5 overall with at least 6.0 in each component (reading, writing, listening and speaking) or equivalent. The test must be dated within two years of the start date of the course in order to be valid.

<h2 class="heading">Funding on offer</h2>

<p>This opportunity is funded by the National Institute of Health Research (NIHR). The scholarship will attract an annual tax-free stipend of &pound;19,237 for year one, and this will increase each year for up to 3 years subject to satisfactory progress. Academic fees will also be paid at the UK fee rate. This scholarship opportunity is for Full-time study (duration: 3 years). The award will be made for one year in the first instance and renewable for a further period of up to two years, subject to satisfactory academic progress.<br /> &nbsp;</p>

<h2 class="heading">Contact details</h2>

<p>For informal enquiries regarding this project please contact Dr Jonathan Benn: <a href=""></a></p> <p>For further information about the admissions process please contact the Admissions Team at: <a href=""></a></p>