|Year : 2018 | Volume
| Issue : 2 | Page : 66-70
Digital medicine scoping: current state and future directions
Neil J Sebire, Shankar Sridharan, Ward Priestman
Digital Research, Informatics and Virtual Environment Unit, Great Ormond Street Hospital, London, UK
|Date of Web Publication||23-Aug-2018|
Neil J Sebire
Great Ormond Street Hospital, London
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Sebire NJ, Sridharan S, Priestman W. Digital medicine scoping: current state and future directions. Digit Med 2018;4:66-70
| Introduction|| |
Technological advances have revolutionized several industries, including previous business models, with disruption from outside the core industry players in many cases., Typical examples are Uber and Amazon, which have significantly changed consumer, and hence supplier, behavior. Healthcare, while at the cutting edge of technology in some areas such as magnetic resonance and positron emission tomography imaging, has not yet been disrupted by the impact of digital technologies (“Digital”). Indeed, in the United Kingdom (UK), the National Health Service (NHS) has remained the single largest buyer of fax machines despite this technology being antiquated in other sectors and countries. Furthermore, many “digital strategies” in health-care largely represent incremental information technology (IT) upgrades, rather than cohesive strategic approaches to maximizing opportunities of future technology and alignment with changes in other industries.
| Categories of Digital Impact in Healthcare|| |
Of course, technology will ultimately transform the entire health-care industry, but at present, broadly, there are four main categories of health-care activity in which digital will impact in short-to-medium term. First, efficiency – using digital tools to do tasks currently performed but in a more efficient manner. An example would be the use of a mobile electronic recording of nursing observations or use of monitors with direct interfaces to patient records. Second, safety – by using digital tools to prevent adverse events and improve patient safety. An example would include automated prescribing guideline adherence systems and recordable operating room checklists. Third, the use of digital to augment the capabilities of existing health-care staff. An example would be the use of clinical decision support tools to suggest investigations or management protocols for particular patient presentations or diagnoses. Finally, there is the ability for digital to perform tasks which simply are not possible for humans carry out. Examples include most artificial intelligence (AI)-based and machine learning (ML)-based tasks such as high-throughput imaging screening and analysis of large datasets, such as genomic data.
While all of these areas provide potential benefit for patients, and also have significant potential for organizational and structural disruption, the vast majority of activity to date has been focused on the former three elements. However, the possibility to deliver exponential change to clinical practice will be predicated on progress in the last area, in which professionals and patients alike embrace the concepts of medical decision-making performed increasingly by digital “intelligent systems.”
| Examples of Potentially Disruptive Healthcare Digital Technologies|| |
A large number of digital approaches are in development and the areas listed below are intended to represent nonexhaustive examples of ways that digital will impact clinical practice (references provided are for illustrative purposes only, and several similar solutions from a range of providers may be available).
With the digitization of image-based investigations that were previously analog, the ability for intelligent systems to provide clinical augmentation, and ultimately diagnosis, is possible. In recent years, almost all radiology departments in the developed world have become digital, and computerized image analysis for chest X-ray and computed tomography screening is well-described in clinical practice. Other specialties are less advanced regarding full-digitization, but it is clear that all image-based investigations will ultimately become digitized, and hence, the data potentially machine learnable (for example, digital pathology is developing as a specific field). Furthermore, once imaging investigations are digitized, and hence simply represent data points, the ability for “intelligent systems” to integrate such data with clinical findings, laboratory investigations, and complex data sets such as “OMICs,” becomes possible, allowing for personalized evaluations and novel observations to be detected. Novel imaging approaches are being developed, and proof of principle of systems for providing diagnostic augmentation are already well described [Figure 1].
|Figure 1: Example of micro computed tomography image of a human early gestation fetus. With such technology resolution of up to 5um can be achieved, with potential for machine learning applications to help identify small anatomical abnormalities (Image courtesy of Dr. O Arthurs, Great Ormond Street Hospital, UK)|
Click here to view
Knowledge search and integration
The NCBI PubMed database contains >27,000,000 scientific health-care publications, with vastly more available online from other sources if abstracts and conference proceedings are included., This volume of scientific data is impossible for any health-care professional to read, even restricted to the available literature in their field, or to be able to search, retrieve and connect features across multiple publications [Figure 2]. Given that, on an average, it takes >15 years for published data to be translated into changed clinical practice by the majority of doctors, simply the ability to rapidly and intelligently make such information available to health-care providers would be transformational. While free-text search engines allow identification of references, evaluation of their content, context, relevance, and connectedness, remain the responsibility of the practitioner. This area is clearly common across industries and requires intelligent natural language processing (NLP) with an understanding of language and concepts rather than simply identification of text, proximity, etc., and is proving challenging. However, several examples have been described using AI systems to provide such information, although, at present, extensive human curation is required. Indeed, the term “Literome” has been suggested as a concept to encompass all the literature related to a particular search, field, or disease.
Sensors and automated physiological data capture
The availability of an increasing range of sensors on consumer devices has expanded, with concomitant increased availability and functionality of health-care-specific sensors. Many of these can capture both clinical observations that would previously have been routinely recorded manually, such as heart rate and blood pressure, in addition to a range of other activities not previously recorded in routine clinical practice, such as micromovements, skin impedance, and geolocation. This technology will lead to the routine availability of substantial volumes of “additional” physiological data, which is not currently part of clinical care, raising both opportunities for development of novel “digital biomarkers” but also raising issues for health-care systems regarding such data capture, storage, and interpretation in the appropriate clinical context.
Augmented, mixed and virtual reality
The ability to present clinically relevant information in novel ways, through augmented reality, mixed reality, and virtual reality systems opens the possibility to allow health-care professionals to interact with patients and their data in previously impossible ways. Examples include apparent three-dimensional representations of imaging datasets, and superimposition of imaging or other data into a surgical operating field. Furthermore, these technologies facilitate entirely new modes of education and learning for both patients and health-care staff, either disease or procedure-specific.
Advances in computational methodologies have resulted in digital simulations, or models, of a range of physiological and pathological processes. Such simulation-based studies have been used for accelerated drug discovery, pharmacokinetics, tumor growth and blood flow modeling, cardiovascular dynamics, and even system-based health economic simulation., The availability of increasingly detailed datasets, in conjunction with scalable, compute capability, will allow evolving advanced simulations, with the ultimate aim of simulating entire human physiology and functions, from which modeling of disease states and therapies can be derived [Figure 3].
The current use of robotics in healthcare is relatively rudimentary, primarily being used for manual tasks such as portering or drug dispensing, and for aiding manual dexterity in surgical procedures. However, with advances in NLP, AI, and ML exponentially more health-care tasks are likely to be augmented or replaced by robots/virtual entities, such as, initially, reception, triage, and standard care advice, but ultimately, the majority of clinical diagnosis and management decisions. The patient acceptance of such major changes to healthcare are likely to be related to the parallel integration of robots and virtual assistants into other areas of consumer and daily life, and this area is often associated with the greatest emotional resistance both by the public and health-care staff.
| Current Challenges to Maximising Digital Health|| |
Paradoxically, the health-care systems themselves remain one of the major barriers to rapid integration of the benefits of digital. Fundamentally, in order for all aspects of digital healthcare to work, existing patient level health data should be available, either for care integration of technologies or research and development of digital health tools. Many health systems worldwide, even in developed countries, continue to rely on paper-based systems of patient records and even systems with more widespread electronic patient record (EPR) implementation may be limited to secondary use of such data, since EPR systems may be proprietary, with no common data format or tools to maximize further data use. There is however, now increasing interest in developing common data interchange, storage, and use standards, such as FHIR and openEHR, which will drive interoperability of systems., Nevertheless, since patient-level data are required for such development, a significant piece of work is required to engage patients and families around safe, secure, and appropriate use of their health data.
In addition, partly due to these underlying systems, many current health-care staff are poorly skilled to implement novel digital technologies and a recent report in the UK highlighted the need for rapid development of clinical/health informaticians, who are trained and competent in both digital and healthcare understanding. As such, IT, EPR platforms and the integration of novel technologies should be seen as “core business” for hospitals, rather than the role of IT as a “support” service for clinical care.
| Health Data Infrastructure|| |
To derive benefit from current routine and future health-care data to improve outcome for patients, platforms are required which allow secure, scalable storage of data for secondary or research use, in addition to robust mechanisms for deidentification of datasets. Furthermore, extensive governance arrangements are required to include control of activity, data access, auditability, and data quality. Nevertheless, the potential benefit of leveraging health-care data has been recognized by two recent reports on behalf of the UK Government,, and significant work is ongoing to formalize elements of health data use for public good-based around the Health Data Research UK infrastructure, in addition to public consultations and developments by NHS Digital.
| Implementing Digital in Healthcare|| |
Although, the elements are in place for potential massive transformation in health-care delivery based on digital technologies, as described above, to derive benefit for patients at scale, the challenge will be driving implementation at the level of individual institutions. The importance of senior, executive-level, clinical staff, such as Chief Clinical Information Officer and Chief Nursing Information Officer to ensure clinical engagement has increasingly been recognized. However, widespread change will require the development of clinical care staff with an openness to new technology, and an understanding of technical aspects require to drive implementation. The UK has recently established the Faculty of Clinical Informatics to drive this development, promoting technological understanding, and expertize as an increasingly important component of healthcare.
| Conclusion|| |
Developments in digital technologies are reaching the point of providing opportunities for significant transformation of healthcare with widespread potential for patient benefit. However, to maximize such opportunities, major new strategic and infrastructure models are required for health-care organization, health professional training and expertize, and models of direct clinical care.
| References|| |
van Ginneken B. Fifty years of computer analysis in chest imaging: Rule-based, machine learning, deep learning. Radiol Phys Technol 2017;10:23-32.
Williams BJ, Bottoms D, Treanor D. Future-proofing pathology: The case for clinical adoption of digital pathology. J Clin Pathol 2017;70:1010-8.
Lu Z. PubMed and beyond: A survey of web tools for searching biomedical literature. Database (Oxford) 2011;2011:baq036.
Morris ZS, Wooding S, Grant J. The answer is 17 years, what is the question: Understanding time lags in translational research. J R Soc Med 2011;104:510-20.
Poon H, Quirk C, DeZiel C, Heckerman D. Literome: PubMed-scale genomic knowledge base in the cloud. Bioinformatics 2014;30:2840-2.
Majumder S, Mondal T, Deen MJ. Wearable sensors for remote health monitoring. Sensors (Basel) 2017;17. pii: E130.
Luo C, Jia J, Gong Y, Wang Z, Fu Q, Pan C, et al.
Highly sensitive, durable, and multifunctional sensor inspired by a spider. ACS Appl Mater Interfaces 2017;9:19955-62.
Okamoto T, Onda S, Yasuda J, Yanaga K, Suzuki N, Hattori A, et al.
Navigation surgery using an augmented reality for pancreatectomy. Dig Surg 2015;32:117-23.
Alaker M, Wynn GR, Arulampalam T. Virtual reality training in laparoscopic surgery: A systematic review & meta-analysis. Int J Surg 2016;29:85-94.
Shinbane JS, Saxon LA. Virtual medicine: Utilization of the advanced cardiac imaging patient avatar for procedural planning and facilitation. J Cardiovasc Comput Tomogr 2018;12:16-27.
Zafari Z, Bryan S, Sin DD, Conte T, Khakban R, Sadatsafavi M, et al.
A systematic review of health economics simulation models of chronic obstructive pulmonary disease. Value Health 2017;20:152-62.
Viceconti M, Hunter P. The virtual physiological human: Ten years after. Annu Rev Biomed Eng 2016;18:103-23.
Ross SB, Downs D, Saeed SM, Dolce JK, Rosemurgy AS. Robotics in surgery: Is a robot necessary? For what? Minerva Chir 2017;72:61-70.
Bosca D, Moner D, Maldonado JA, Robles M. Combining archetypes with fast health interoperability resources in future-proof health information systems. Stud Health Technol Inform 2015;210:180-4.
Lamas E, Barh A, Brown D, Jaulent MC. Ethical, legal and social issues related to the health data-warehouses: Re-using health data in the research and public health research. Stud Health Technol Inform 2015;210:719-23.
Kannry J, Sengstack P, Thyvalikakath TP, Poikonen J, Middleton B, Payne T, et al.
The chief clinical informatics officer (CCIO): AMIA task force report on CCIO knowledge, education, and skillset requirements. Appl Clin Inform 2016;7:143-76.
[Figure 1], [Figure 2], [Figure 3]