Skip main navigation

Digital Transformation: How Can Big Data Be Used?

In this second part of our look at digital transformation, Dr Dan Bendel describes how data can be used to answer more complex research questions.

In this second article on digital transformation, Dr Dan Bendel now looks at ways in which data from EHRS can be used to answer research questions and ultimately improve perioperative care.

Clinicians as Data Architects

To fully harness the potential of EHR data, clinicians must understand the principles of data pipeline design. EHRs store clinical data as “structured data”, meaning that it ultimately gets stored in labelled tables. This makes this data amenable to automatic retrieval later. Proper data pipeline management (i.e. paying attention to how you record your data, how it is stored and whether this yields reliable results later on) ensures the accuracy and relevance of the data, facilitating its use in governance, quality improvement, and research.

From structured data to Computable Phenotypes

Structured EHR data can imply important clinical entities that are not necessarily documented. By defining logical expressions based on available data, clinicians can design “computable phenotypes”. This further enables the ability to identify specific patient cohorts and correlate them with clinical outcomes. This technology means that you could explore the relationship between the choice of anaesthetic and postoperative nausea and vomiting, for example.

Multi-Centre Registries

This isn’t just happening on a local scale. Automated, multicentre perioperative registries, such as the Multicentre Perioperative Outcomes Group (MPOG) in the USA, represent a significant advancement in how perioperative data is collected, stored and presented [1]. These registries enhance data reliability and support large-scale research, accelerating and guiding future healthcare strategies.

This raises a question: will every hospital on the planet one day contribute its data to a global perioperative registry?

UK Perspective: “Data Saves Lives”

In 2022, Ben Goldacre’s review for the Department of Health and Social Care highlighted the importance of future, centralized NHS data platforms [2]. Recommendations included the creation of national trusted research environments (TREs) to improve data access and privacy. These TREs would enable faster, more secure data sharing and support collaborative research [3]. The ongoing digitalization of NHS trusts aims to leverage EHR data for better healthcare outcomes, aligning with the goals outlined in the “Data Saves Lives” policy document.

AI and Machine Learning in Perioperative Medicine

AI and machine learning (ML – a subset of AI) offer powerful, if experimental, tools for analyzing more complicated healthcare data [4]. At the time of writing, AI is sufficiently sophisticated as to arrive at conclusions about clinical trends and patterns, paving the way for AI-powered patient alerts and clinical decision support [5]. The ensuing algorithms have a role in perioperative practice, and have been found to maintain physiological parameters in a closed-loop manner within tighter physiological limits than usual manual means. Examples include weaning patients off ventilators and vasopressor administration in sepsis on intensive care [6],[7],[8],[9]. In theatre, ML algorithms trained to analyse invasive arterial line traces are now able to predict hypotensive events fifteen minutes in advance of them happening [10]. Working experimental models have been able to predict the likelihood of AKI, reintubation and mortality from EHR data available at the end of surgery alone [11].
However, the use of AI in healthcare raises concerns about transparency, regulation, and safety. Neural networks, the foundation of most AI systems, operate as “black boxes” with logic rules that are not easily understood by humans. For example, in 2018 a neural network was trained to link target-controlled infusion rates of propofol and remifentanil to the bispectral index (BIS) in 231 individuals. Their model performed well against generally accepted metrics of drug performance, and without any prior knowledge of propofol or remifentanil pharmacokinetics, or the Schinder or Minto models used in the study [12].
This lack of transparency and the absence of regulatory standards pose challenges for integrating AI into clinical practice. Ensuring patient safety and maintaining clinical governance will require rigorous testing and regulation of AI algorithms, similar to the approval process for new drugs [13].

A local example – the Clinical Research Informatics Unit (CRIU)

Despite these challenges, AI has already demonstrated value within the NHS. The Clinical Research Informatics Unit (CRIU) at UCLH oversee several digital healthcare programs, including the Experimental Medicine Applications Platform (EMAP) [14]. This platform mirrors real-time healthcare data, allowing for the safe testing of experimental processes. EMAP uses machine learning for predictive analytics, such as forecasting ICU bed demand and reporting on antimicrobial resistance patterns. CRIU has also developed CogStack, a natural language processing algorithm able to interrogate free-text documents stored within EMAP, such as clinic letters and GP summaries [15]. It can extract and report on key clinical concepts from these documents, and has already contributed to clinical trials [16]. By maintaining data within the NHS and using standardized international data definitions, EMAP maintains data privacy whilst facilitating collaboration with other institutions [17].

To Conclude

Digital innovation in perioperative medicine, driven by EHR adoption, enhances operational performance, patient safety, and research capabilities.

The integration of telemedicine, wearable technology, and AI further improves clinical care by enabling remote monitoring and predictive analytics. However, the rapid pace of digital transformation necessitates upskilling and ongoing regulation to fully leverage and manage these technologies. Ensuring that clinicians are equipped to handle the complexities of digital healthcare is mission-critical if we are to fully realise its potential in a manner that is safe.

References (Continued)

  1. Multicentre Perioperative Outcomes Group. MPOG [Internet]. 2023 [cited 2023 Jun 7]. Available from:
  2. Goldacre B. Better, Broader, Safer: Using Health Data for Research and Analysis. 2022.
  3. The Department of Heath and Social Care. Data Saves Lives [Internet]. [cited 2023 Jun 15]. Available from:
  4. Gambus P, Shafer SL. Artificial intelligence for everyone. Vol. 128, Anesthesiology. Lippincott Williams and Wilkins; 2018. p. 431–3.
  5. Bennett CC, Hauser K. Artificial intelligence framework for simulating clinical decision-making: A Markov decision process approach. Artif Intell Med. 2013 Jan;57(1):9–19.
  6. Rose L, Schultz MJ, Cardwell CR, Jouvet P, McAuley DF, Blackwood B. Automated versus non-automated weaning for reducing the duration of mechanical ventilation for critically ill adults and children. Cochrane Database of Systematic Reviews. 2014 Jun 10;2018(12).
  7. Brogi E, Cyr S, Kazan R, Giunta F, Hemmerling TM. Clinical Performance and Safety of Closed-Loop Systems: A Systematic Review and Meta-analysis of Randomized Controlled Trials. Anesth Analg. 2017 Feb;124(2):446–55.
  8. Mathis MR, Kheterpal S, Najarian K. Artificial Intelligence for Anesthesia: What the Practicing Clinician Needs to Know. Anesthesiology. 2018 Oct 1;129(4):619–22.
  9. Komorowski M, Celi LA, Badawi O, Gordon AC, Faisal AA. The Artificial Intelligence Clinician learns optimal treatment strategies for sepsis in intensive care. Nat Med. 2018 Nov 22;24(11):1716–20.
  10. Sidiropoulou T, Tsoumpa M, Griva P, Galarioti V, Matsota P. Prediction and Prevention of Intraoperative Hypotension with the Hypotension Prediction Index: A Narrative Review. Vol. 11, Journal of Clinical Medicine. MDPI; 2022.
  11. Hofer IS, Lee C, Gabel E, Baldi P, Cannesson M. Development and validation of a deep neural network model to predict postoperative mortality, acute kidney injury, and reintubation using a single feature set. NPJ Digit Med. 2020 Dec 1;3(1).
  12. Lee HC, Ryu HG, Chung EJ, Jung CW. Prediction of Bispectral Index during Target-controlled Infusion of Propofol and Remifentanil. Anesthesiology. 2018 Mar 1;128(3):492–501.
  13. Panchagnula U, Shanmugam M, Rao BM. Digital future in perioperative medicine: Are we there yet? J Anaesthesiol Clin Pharmacol. 2019 Jul-Sep;35(3):292-294.
  14. NIHR UCLH Biomedical Research Centre. Clinical and Research Informatics Unit (CRIU) [Internet]. 2024 [cited 2024 May 26]. Available from:
  15. Jackson R, Kartoglu I, Stringer C, Gorrell G, Roberts A, Song X, et al. CogStack – experiences of deploying integrated information retrieval and extraction services in a large National Health Service Foundation Trust hospital. BMC Med Inform Decis Mak. 2018 Dec 25;18(1):47.
  16. Bean DM, Kraljevic Z, Searle T, Bendayan R, Kevin O, Pickles A, et al. Angiotensin‐converting enzyme inhibitors and angiotensin II receptor blockers are not associated with severe COVID‐19 infection in a multi‐site UK acute hospital trust. Eur J Heart Fail. 2020 Jun 7;22(6):967–74.
  17. Observational Health Data Sciences and Informatics. The Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) [Internet]. 2024 [cited 2024 May 26]. Available from:
This article is from the free online

Perioperative Medicine in Action

Created by
FutureLearn - Learning For Life

Reach your personal and professional goals

Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.

Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.

Start Learning now