From the library of What’s next: archives from From the editor

Oxford University Hospital boosts its health digitisation plans

Oxford University Hospital's ongoing programme of investment in digital services and infrastructure is ‘Go Digital’. It has ambitious plans to accelerate the opportunities that digital technology offers. This is in line with the vision of the NHS to be ‘paper-free’ and for patient records to be held electronically and accessible across different systems. Clinical speech recognition has done much to boost this program, reduce costs and free clinicians to care
By

 

Going Digital

Oxford University Hospital NHS Foundation Trust’ (OUH) ‘Go Digital’ health digitisation programme is a journey, not a ‘big bang’. Key to ‘Go Digital’ is the rollout and adoption by clinicians of the ‘Go Digital’ platform – the Cerner Millennium electronic patient record (EPR). OUH’ aim is to deliver information to clinical teams based on real-time data and enable them to share that information with colleagues across different record-keeping systems and to underpin high quality care and improve communication with patients.

How clinical speech recognition found its voice

Amidst the ambitious aims of ‘Go Digital’ are the day to day challenges of delivering health services in response to growing demand and constrained budgets. In 2017, one well-performing department at the OUH was averaging a 12 day turnaround of clinic letters to General Practitioners (GPs) and struggling to meet the clinical commissioning group (CCG) target which as of April 2018 is 5 days. The root cause was a combination of a chronic shortage of administration staff and the complex, costly workflow of  in-house and outsourced transcription used to produce the outpatient clinic letters. Printing and mailing of letters added to delay and cost.

Consultant nephrologist and OUH’s Chief Clinical Information Officer (CCIO), Dr Paul Altmann, piloted and championed the use of Nuance Dragon Medical front-end clinical speech recognition in nephrology within the Cerner Millennium EPR. Using a structured clinic letter template mirroring their legacy system workflow he then shared this with a handful of co-piloteers and quickly realised the potential of Dragon Medical integrated into the EPR to simplify workflow, save clinician time and cut costs associated with clinic letter production.

Building the business case

OUH initiated a 3-month pilot of Dragon Medical One; secure, clinical speech recognition in the cloud. The pilot was across a range of specialties and with a focus on the whole of the nephrology department. The success criteria for the pilot were set out from the start:

  • Achieve adoption of speech recognition by at least 80% of designated users
  • Reduce clinic letter turnaround times
  • Reduce outsourced transcription costs
  • Ensure complete integration with EPR
  • Drive the Go-Digital and paperless strategy

Services speed the change

Throughout the pilot Nuance Professional Services (PS) delivered workflow analysis and one to one training for Dragon Medical One for the nephrology team and the Trust’s own EPR trainers. Once Dragon Medical One licenses were enabled and actively in use by the clinicians, Nuance Client Success Organisation (CSO)constantly monitored the progress of uptake and adoption of the licenses by the clinicians. To support the pilot effort Nuance PS and CSO and OUH project team  worked closely. Together they carried out weekly project reviews to quickly identify and fix any training or process issues. The lessons learned from these weekly meetings further hastened rollout.

Successful pilot and lasting benefits for health digitisation

The success criteria of the pilot were fully achieved including:

  • 100% adoption of Dragon Medical One speech recognition
  • Clinic letter turnaround times reduced from 12 to 3 days
  • Outsourced transcription no longer used i.e. zero cost

Having proven that the transition from transcription and digital dictation workflow to front-end speech recognition is feasible and cost effective, the roll out of Dragon Medical One continues apace and is due to complete in November 2019. OUH investment in secure, cloud-based clinical speech recognition will continue to accelerate health digitisation and deliver long lasting benefits to clinicians, patients and the organisation as it continues to roll out across the whole of the Trust’s 8000 clinicians.

 

Boost your hospital digitisation plan and free your clinicians to care

Discover all the hard and soft benefits delivered by secure, cloud-based clinical speech recognition at Oxford University Hospital

Learn more

Tags: , , , , ,

5 top tips for rolling out successful clinical speech projects

When Homerton University Hospital NHS Foundation Trust made the decision to transition to speech recognition, replacing all previous methods of transcribing letters, it was imperative that everyone was on board.
By
NHS Digital must do its homework on clinical documentation

1. Re-imagine the process

After gaining board support for the project, the first thing the transformation team at Homerton did to prepare for the roll-out of Dragon Medical One was to reflect on current clinical workflows. A process mapping exercise ‘re-imagined’ the process and revealed unnecessary steps to current workflows which could be cut out by deploying speech recognition. A new electronic workflow map was designed, a pilot programme demonstrated proof of concept and the benefits of making the change.

2. Choose the right clinical leader

The choice of clinical lead can have a significant impact on how other clinicians engage with a project. The Transformation Team carried out interviews to ensure the clinical lead was pro-change, energetic, had trust-wide networks and was someone who people responded to positively. Dragon Medical was rolled out to the clinical lead’s department first, then remaining stakeholders were mapped out into three tranches:

  • Engaged clinicians: those traditionally in favour of technology who would support the project.
  • Most services: those people who accept change
  • Dis-engaged clinicians: the few clinicians who had workflows which were different, traditionally needed more support or had responded negatively to previous change implementation.

3. Clinician engagement

When it comes to clinician engagement in new technology “there will always be people who are positive about new solutions and embrace change and there will always be people who are more skeptical,” says Katherine Adams, Transformation Manager and ED Senior Sister at NHS Homerton. In our recent webinar, Re-imagining outpatient services, she talks about some of the approaches the Trust had taken when it came to rolling out Dragon Medical.

4. Invest in training, support and communication to ensure clinicians feel safe and secure

Nuance worked with the Trust throughout the roll-out of Dragon Medical by holding classroom training sessions and giving practice exercises to reinforce learning.

Positive Change leaders provided floor support and a working group met regularly throughout the project. Transformation team members attended department meetings to provide updates and address any questions or queries.

Communication on how the project was progressing was reported through regular staff emails, blog posts and hospital magazine articles. Messages of support were sent from the trust medical director, operational director and both the clinical information systems and IT teams.

5. One to one

Despite the communication, support and training put in place, some clinicians remained reluctant to engage with the changes. Katherine had to ensure she devoted some one-to-one time with these clinicians so she could address their concerns and point out the benefits of the project. However, she says: “It was worth spending time with those people to bring them on board.”

Winning Results

Following the roll out of Dragon Medical across all adult services, letter turnaround time has reduced from 17.7 days to just 2.2 day with, on average, 80 per cent going out within 24 hours. In total 40,000 letters now go out per month.

The Trust has been able to save a third of its medical secretariat budget and reduced its outsourcing costs by £180,000 per annum.

However, the most significant impact of this change in workflow has been on the Trust’s patients. Clinical documentation and information is shared between the trust, GPs and patients much faster, which means any treatment a patient requires can be started earlier than before.

Download the NHS Homerton case study here

Discover how clinicians, admin staff and operational management at the Trust are championing the change to Dragon Medical One secure cloud-based speech recognition.

Learn more

Tags: , , ,

Five steps toward NHS cyber security compliance

There are a bewildering number of guidelines and rules when it comes to meeting NHS cyber security, safety, privacy and risk management for any organisation working in the UK healthcare sector. For example, the documentation alone required to set up as a software vendor to the NHS can be daunting. Depending on the size of your company and the resources available to you, some of these certifications may seem too complex to put in place. However, if you take them one at a time, getting the right certifications is important and will pay off in the long run. Here are my top five tips for healthcare software providers:
By
Black key locked in to represent NHS cyber security compliance

1. Start as you mean to go on

Make sure you have clear company policy documents covering staff and employment practice, and that you can prove that the policies are working – this gets more important as you ascend the heights of Information Governance (IG) compliance.

2. Get the basics right

Register with the Information Commissioners Office where there is lots of information helping you get your GDPR and Data Processing agreements and policies in place. It is important to conduct Privacy Impact Assessments for your software externally and your processes internally. Make sure your staff are regularly trained on Information Governance and you can prove it. Also make sure you are registered on the Organisation Data Service with your primary contracting entity. It is also a good idea to sign up for Cyber Essentials (Plus)

3. Make sure you comply with DCB0129

This lesser known guideline kicks in when you start processing patient data, or you are involved in decision support or telehealth. This involves performing Clinical Risk Management on all changes and new features in your software. It is a development task resulting in a Safety Case document showing the risk analysis before and after changes and should be released in line with your regular release notes.

4. Comply with Data Security and Protection Toolkit

Complying a data security and protection toolkit is a more involved process and one which starts you on the road to having ISO27001. This online questionnaire requires you to evidence all processes and procedures relating to Data Security and protection. If you have done the above properly then you should have these processes in place such as internal governance policies, staff contracts and training and physical and cyber security. Most NHS Trusts will require this as the basic standard for working with patient data.

5. Meet ISO27001

This usually satisfies most security related queried from the NHS. Depending on how organised you have been in the previous sections this could be a relatively simple certification. Alternatively, it can be a time consuming task if you are a large, disparate organisation. Scope here is everything – define this well and save lots of time. In my experience it is easier for smaller companies to achieve this if they have the processes in place already and it is economically viable. This is especially relevant if you are hosting a solution into the NHS or if you provide services from abroad. You must be externally certified for all related processes and IG policies as well as security management systems, physical security, business continuity, incident reporting and so on. My advice is to create a definitive security document encompassing all the certifications here for each client. They will never doubt your security again.

Nuance Dragon Medical One clinical speech recognition meets NHS cyber security compliance

Read more about our secure clinical speech recognition solutions in the cloud for the NHS

Learn more

Tags: , , ,

AI needs quality NHS data to succeed

I recently revisited the Annual Report of the Chief Medical Officer, published at the end of last year, in which Professor Dame Sally Davies takes “an aspirational view of what health could and should look like in 2040”. In doing so, she provides some hope about the future of health in the UK.
By
Male doctor pointing at different medical features

The vision of personalised data

The Annual Report of the Chief Medical Officer suggests we will “evolve from Electronic Health Records to an individualised ‘Electronic Health Engine’ that integrates high dimensional data about the individual, including social and economic determinants of health, behavioural risks, biomedical, genomic and citizen-generated data, to generate real time dynamic risk trajectories”. This vision places personalised data at the heart of the health service. Data is a central theme throughout the report – Dame Sally Davies dedicates four of the 15 chapters to its role in the future of healthcare, including in artificial intelligence.

AI is founded on high quality clinical information

It is widely acknowledged that artificial intelligence (AI) holds huge potential for healthcare. The Annual Report highlights Dr Dominic King, clinical lead at DeepMind, who has stated that AI is capable of “unlocking the full potential of other promising advanced technologies being developed in medicine”. He goes on to say that, through AI, data and technology can be used to “achieve a much more precise grasp of potential treatment needs, both at the patient and population level”. If the full potential of AI in healthcare is to be reached, however, it will require a foundation of high-quality clinical information. The measurement and recording of this information must, therefore, become an automatic feature of interactions on a patient’s healthcare journey.

Speech to text saves clinician time and improves the quality of data recorded 

In a previous blog, I highlighted that up to 50 per cent of clinician time can be spent on documentation, while only around 13 per cent of their working time is spent with patients. I have also highlighted that speaking is up to three times faster than typing, making speech recognition software a key method of releasing more clinician time away from administrative tasks. However, the health system will only benefit from this if the information going into the Electronic Patient Record, or the ‘Electronic Health Engine’, is accurate and good quality information. The best way of ensuring this is to make the recording of this information a seamless and natural by-product of clinician-patient engagement. Speech to text has been shown to improve the quality of information recorded during consultations.

Making a difference at the frontline 

There are several examples where speech recognition software has helped in this way. In the outpatient department, clinicians using it has led to turnaround time for letter to GPs and patients dropping from weeks to just two to five days at the Homerton University Hospital NHS Trust. This has also been replicated at Oxford Universities Hospitals NHS Foundation Trust. We have found similar outcomes in a community physiotherapy service, a mental health service and histopathology service. In the latter, a backlog of 600 pathology cases was removed and the organisation is now exceeding the national turnaround target. In a fast-moving, unpredictable emergency department, using speech recognition software has resulted in a saving of up to 40 per cent of documentation time.

So, while there are a number of key technologies that will enable Dame Sally Davies’ 2040 vision, tools to record personal data in an electronic, codifiable format are essential. Speech recognition software can deliver that, and I believe it will be the primary tool for recording healthcare interactions well before 2040.

How we are helping the NHS

Learn how we are helping the NHS capture high quality clinical information accurately and more quickly

Learn more

Tags: , , ,

AI and the medical consultation of the future

What if the medical consultation of the future used a virtual medical assistant based on AI? What if we could allow doctors to devote themselves completely to their patients and free them from the tedious task of entering the data into the electronic patient record?
By
Male nurse and boy using stethoscope
AI and the transformation of the patient-doctor relationship

This week (February 11-15, 2019) is HIMSS 2019 in Orlando USA where health professionals and practitioners from around the world gather. For the first time, visitors to the Nuance stand will be able to experience for themselves how AI can transform the patient-doctor relationship.

What will this experience feel like?

An increased interaction between the doctor and patient, where the conversation comes to the fore, the search for information is simple, where the key data from the conversation throughout the consultation is sensed and then automatically captured directly into the electronic patient record.

AI and Ambient Clinical Intelligence

Ambient Clinical Intelligence guides the doctor-patient encounter with assisted workflows, automation of tasks and knowledge, as well as specialised equipment for ambient sound detection. With Ambient Clinical Intelligence clinicians are able to focus on their patient rather than on a screen. The documentation of the patient’s file is carried out automatically and doctors benefit from automated clinical advice.

The medical consultation of the future

For the patient, it is about having time, being listened to and having the full attention of the doctor and vice versa. We can probably all relate an example of a consultation where the doctor was in a hurry and as frustrating for the doctor as the patient.

The promise of digital technologies such as AI and Ambient Clinical Intelligence are expected to create a new space for interaction and trust. Paradoxically, we could say that AI puts the humanity back into medicine!

Learn more about the medical consultation of the future

Discover how AI-based Nuance solutions are already transforming clincians' lives and relationships with their patients

Learn more

Tags: , , ,

UK Health Tech Predictions 2019

Nuance healthcare international CCIO Simon Wallace shares his thoughts on health tech innovations and expectations for 2019
By
UK health tech predictions 2019
What would be your new year’s resolution for the health tech industry for 2019?

This year, the NHS must demonstrate how it can encourage a culture shift to ensure technology is properly being used to boost efficiency, improve patient care and reduce stress and burnout seen across the healthcare profession. To achieve this, it is important that budget allocated to digital health is utilised in this way, and not clawed back to fund other reactive needs, such as winter pressure. Underpinned by secure advances in cloud computing, global digital exemplar Trusts – and their fast followers – must demonstrate they are the world leaders in harnessing digital technology to improve the delivery of patient care.

What are you most excited to see in health tech this year?

I’m most excited by the potential of achieving a single summary view of each patient’s healthcare record – with key pertinent summary details for healthcare professionals to see – whatever part of the journey the patient is on. There have been several attempts at achieving this in the past, but we have the technology available today to make it happen.

Alongside this, it will be easier for patients to access and consult with their healthcare practitioner – for example, via video consultations – so they can get access to care and treatment faster, at a time of their convenience and at less expense.

All of this requires high quality clinical documentation, and that’s where technology such as speech recognition comes in – enabling clinicians to compile records using just their voice to capture the patient story completely and accurately at the point of care.

What are some of the biggest challenges facing health tech this year?

The NHS will always be under pressure. We have a growing and aging population that’s demanding more from clinicians than ever before. Therefore, we need a way to maintain the enthusiasm of clinicians, in spite of the stressed environment.

Achieving this won’t be a simple process, and much of it will hinge on ensuring promised budgets for digital technology remain. Should we manage to do so, change management will be the next key step. The introduction of digital technologies will always require training and support – helping Trusts ensure the technology and new approaches are embedded and adopted. This will require the organisation to accept that time and resources must be dedicated to it.

With such backing the NHS will be able to increase the adoption of electronic patient records, integrate patient data in a meaningful way and link with social care systems to provide a complete patient overview – helping clinicians provide a better service at the point of care.

What technologies do you see as having the biggest impact this year?

The cloud will transform patient services, with its scalability, ability to reduce expenditure by not having to invest in additional hardware or recruit expensive technical resources to run the software day-to-day. With costs cut on technical support and management overheads – and software continuously updating – Trusts can look away from the management of technology and focus on the delivery of patient services.

This, alongside the increasing use of artificial intelligence (AI), should reduce the burden of administration and support clinical decision-making.

How do you think health tech has changed recently?

Health tech hasn’t significantly ‘changed’ in the last year but by necessity and time, adoption of digital technology has grown – as has the adoption of AI, which is becoming more pervasive and will expand to more often support their clinical decision freeing clinicians to focus on patient care.

Read how Homerton are using Nuance healthcare solutions to make clinicians lives easier in 2019

Dragon Medical One speech recognition in the Cloud contribution to UK healthcare

Learn more

Tags: , ,

Re-imagining clinical documentation in outpatient services

NHS Homerton's investment in Dragon Medical One has enabled clinicians and medical secretaries to re-visit outpatient workflows and processes and accompanying clinical documentation. They have re-designed them for long-term efficiency and to free-up precious administration resources to refocus them on patient-centred activities.
By
Out with the old

Clinicians in outpatients had been reliant on handwritten notes and using medical secretaries to type and edit letters. Letters could take up to 17 working days to reach patients. Budget constraints meant that recruiting more medical secretaries or using agency staff was not an option. An alternative method of quickly and accurately producing clinical documentation was required.

The transformation team implemented Nuance Dragon Medical One
cloud-based speech recognition integrated into the Cerner Millennium electronic patient record (EPR). Clinicians now use their voice to capture the patient consultation more naturally, efficiently and on their own terms.

Paul Adams, Head of Clinical Information Systems said: “The roll out of Dragon Medical One to our clinicians was quick and simple and limited only to the speed at which we could introduce clinicians and their teams to the new ways of paperless working.”

In with the new

Within weeks, turnaround time of clinic letters dramatically reduced from 17 days to just two. The trust estimates it is saving in excess of £150,000 per year and bank and agency costs have been reduced culminating in a one third reduction in spend on the medical secretariat

Furthermore, during outpatient clinics, clinicians can now enter their own notes into the electronic patient record at the point of care and these notes can be immediately sent to the GP. Clinicians create clinic letters and give them to their patients before the patient has left the clinic.

Outpatient services re-imagined

Resulting in positive change for everyone:

  • Patients receive faster, personalised communication.
  • Medical secretaries can now focus on patient contact rather than typing backlogs.
  • Clinicians know they are providing a better level of care to patients and no longer have to spend extra time after clinics catching up on paperwork.

Dr Rob Fearn, Gastroenterologist, said: “Change always causes some anxiety, but that isn’t a reason not to make the changes. The way to make it work is to really understand the problem that the change is trying to solve. By bringing together clinicians, admin staff and operational management from the Trust I think we’ve been able to do that well.”

 

 

Download the NHS Homerton outpatient services case study

Learn more about Dragon Medical One has reduced transcription costs, sped up clinical correspondence and freed-up time to care.

Learn more

Tags: , , , , ,

Mobilising clinicians in community and mental health

Worcestershire Health and Care NHS Trust is one of seven mental health trusts chosen as a Global Digital Exemplar (GDE); NHS England’s flagship digital initiative, prioritising funding for the most digitally advanced trusts. Worcester’s GDE digital and technology investments have focused on better access to patient records by enabling mobile access to the patient record system so that community and mental health teams can update patient records and other clinical documentation on-the-go and without needing to return to the hospital or clinic.
By
Case study describing how clinical speech recognition supports mobile health workforce in community and mental health

 

Enhancing mobile working

    Recently Worcester used GDE  funding to equip its healthcare workforce with clinical speech recognition to support remote working, reduce clinical documentation workload, eliminate the backlog of reporting associated with detailed patient records and replace legacy, slow analogue dictation workflows with the goal of freeing up healthcare workers to focus on patient care.
     

    Lifting the burden

      Many of the community and mental health teams make extensive notes to capture the patient story and the context of their clients’/patients’ care. These notes are vital in communication with colleagues in multi-disciplinary health and care teams to ensure continuity of care and to meet child protection, medico-legal and other social care requirements.

      With no back-office administration support, many of the team were spending long hours capturing patient records, writing GP letters and other clinical documentation. The results of this were people going home late or producing abbreviate notes which in turn were difficult for others to interpret or caused duplication of effort.
       

      Its personal

        The introduction of clinical speech recognition has boosted mobile working by reducing the burden of paperwork and backlog of administration amongst paediatricians, psychiatrists, community and mental health nurses and

        AHPs working with their patients in the community. For one occupational therapist in particular, clinical speech recognition has changed not just her own working life. With her renewed enthusiasm for technology the effects have ripped down to positively impact on her patients too.
         

        Leading by example

          The success of Worcester’s technology investments for its mobile health workforce has come under the scrutiny of NHS England and NHS Digital. The lessons learned from use of speech recognition within the clinical documentation workflow will be communicated to and showcased for other community and mental health trusts.

           

          Mobilise your clinicians

          Read how Worcestershire Health and Care enabled their community workforce with clinical speech recognition

          Learn more

          Tags: , ,

          What does the transition to SNOMED CT mean for big data in healthcare?

          The NHS is set for a phased transition to SNOMED CT this month, meaning that the public healthcare system will conform to one standardised vocabulary of clinical terminology that consists of over 300,000 medical terms. However, will the integration of a singular coding regime be as simple to implement as the healthcare sector had hoped? This new clinical coding language offers many opportunities for clinicians, but it is worth noting the issues that may also arise.
          By

          The NHS phased transition to SNOMED CT began earlier this month, marking the beginning of a new coding regime, which will facilitate the seamless exchange of coded documents. A standardised vocabulary of clinical terminology intended to be used in Electronic Patient Records (EPR), SNOMED CT contains in excess of 300,000 medical terms that can then be easily recognised across different hospitals, practices, and even countries.

          According to the official SNOMED website it is, “the most comprehensive, multilingual healthcare terminology in the world [and] is already used in more than fifty countries… SNOMED CT can be used to represent clinically relevant information consistently, reliably and comprehensively as an integral part of producing electronic health information.”

          In practical terms, the rollout of SNOMED – or Systematised Nomenclature of Medicine – Clinical Terms – will help NHS practices and hospitals to capture patient records with greater depth and accuracy. The replacement of three existing medical vocabularies with a single coding system will also reduce any possible confusions that may occur and thus improve patient care, particularly when transferring patients.

          SNOMED CT also presents an opportunity to analyse big data in healthcare, with the standardisation of coding making it easier to identify trends and draw conclusions at both local and national levels.

          However, what are the challenges that lie ahead during the implementation?

          SNOMED CT – Opportunities and issues

          As clinical coding becomes standardised across the NHS, there is a clear opportunity for medical professionals at all levels to reduce errors and create richer electronic patient records. But in order to do so, it is vital to use complimentary medical coding products that will make it possible to get the best out of the new system for coding in healthcare.

          A Guardian article about the work of a clinical coder reveals a potential seam for problems, even with new coding: “This time, poor handwriting and conflicting accounts by clinicians make it difficult to determine whether my patient has a malfunctioning tracheostomy or if it’s her artificial voice box playing up.”

          While errors of this sort can be reduced by the digitisation of patient records, how those records are created in digital form is important. For example, if practice GPs dictate these notes to be typed up by secretaries, errors can still occur, with terms being misheard or otherwise recorded erroneously. Medical professionals typing up patient notes themselves meanwhile can also produce errors, as well as taking additional time that could be spent treating patients.

          Digital dictation – The key to accurate clinical coding

          Medical transcription software such as Dragon Medical provides a solution to the problem of inaccurate coding, as well as saving time and resources. Direct voice-to-text dictation with accuracy levels of 99%+ ensures that what is said is what is recorded. As Dragon Medical solutions can be setup to record direct to the EPR, substantial amounts of time can be saved, whether in non-medical staff hours spent typing or time spent by GPs typing the patient notes themselves.

          Dragon Medical solutions can work in tandem with SNOMED CT to revolutionise clinical coding and the EPR. While SNOMED provides the common language with which to describe and categorise patients sessions, Dragon Medical solutions provide the direct intuitive interface and accuracy to enable medical professionals to create rich and detailed narratives.

          In short, use of medical transcription software can help improve the integrity and depth of clinical coding and patient notes, by providing clinicians with the means to transform their own voice into accurate digital notes.

          Find out more about Nuance’s solutions for medical voice transcription enabling better clinical coding.

          Improve clinical coding compliance today!

          Find out how Dragon Medical solutions are easing the pressures of today’s clinicians, by reducing the time needed for patient documentation

          Learn more

          Tags: , , ,

          How I learned my ABCs: The similarities between AI and toddlers

          Artificial intelligence (AI) is quickly transforming decision-making in healthcare. From improving the accuracy and quality of clinical documentation to helping radiologists detect abnormal images to make them high priority, AI is freeing clinicians to focus more of their brain cycles on delivering effective patient care.
          By

          Now, thanks to the impact of deep neural networks (DNN), the application of AI and machine learning to healthcare may finally be reaching a crucial tipping point. But what are neural networks? One of the best ways to understand this is to think about how children learn.

          I’ve been teaching my two-year old about animals, pointing to different ones in a book. It struck me that there are a lot of similarities in the basic elements of animals, yet small children are able to learn and tell them apart. Four legs and a tail— this could be almost any land-dwelling animal. But one has a very long neck while the other has a trunk. These distinguishing characteristics help our brain analyse the information and arrive at the correct conclusions: A giraffe versus an elephant.

          Neural networks are designed to work in much the same way the human brain works. An array of simple algorithmic nodes—like the neurons in a brain—analyse snippets of information and make connections, assembling complex data puzzles to arrive at an answer. The “deep” part refers to the way deep neural networks are organised in many layers, with the intermediate (or “hidden”) layers focused on identifying elemental pieces (or “features”) of the puzzle and then passing what they have learned to deeper layers in the network to develop a more complete understanding of the input and produce a valid output.

          Just like my two-year-old, and all other humans, the network is not born with specific knowledge; it must be trained, like understanding the difference between a giraffe and an elephant noticing one has a big neck and the other has a short one. By feeding the network large amounts of data with known answers, we are effectively “teaching” it how to interpret and understand various inputs— this is also known as “machine learning.” For example, training a DNN to perform medical transcription might involve feeding it billions of lines of spoken narrative and resulting textual output to create a “truth set”—spoken words connected with accurate text. The truth set expands over time as the DNN is subjected to more inputs and the network’s ability to deliver the correct answer becomes more robust. If it gets something wrong, the DNN then must be corrected to reinforce it’s understanding. Like a toddler just learning to identify colours, shapes and animals, the DNN will soon be able to deliver the right answer.

          So how are DNNs changing the way healthcare is practiced? Two areas among many potential applications include clinical documentation improvement (CDI) and radiology image processing. Clinical documentation includes a wide range of inputs, from speech-generated or typed physician notes to labs and medications. Traditionally, CDI involves having domain experts review the documentation to ensure a physician put into documentation an accurate representation of a patient’s condition and diagnosis. However, this approach requires time and resources, and can be disruptive to physician workflow. One approach to automating this process is an arduous, complex processing task that involves capturing and digitising the domain expertise to create a knowledge base, then applying natural language processing technology to then generate a query for the physician in real-time as she is entering her documentation.

          Neural networks improve this process dramatically. Now we can use historical clinical documentation from physicians, including the queries generated by domain experts, to create a truth set for training the neural network. This allows us to skip all the complexity in the middle. The DNN figures that out for itself, based on what it “learned” from the historical truth set. Ultimately, this helps improve documentation by having AI figure out the missing pieces or connections to advise physicians in real time while they’re still charting. What AI is doing here is allowing physicians to focus on patients while the system manages the billing codes, regulatory requirements, quality measures and safety indicators in records.

          DNNs are also changing the game for evaluating visual data, including radiological images. It takes the highly experienced set of eyes of an expert who has studied thousands of similar images to read the subtle clues found there. With neural networks, we can leverage this experience by training the network with thousands of radiological images with known diagnoses. The more images fed through it, the more “experienced” and accurate it becomes, enabling the network to detect the subtle differences between a positive finding and a negative finding. This technology is going to augment the busy workflow of the radiologist and truly amplify their knowledge and productivity by helping them to do things like prioritise the most critical studies. Today when some radiologists read 100 images a day, having AI sift through and spot atypical images to prioritise them first delivers value to physicians and patients who are both looking for the best outcomes.

          The possibilities for neural networks are incredibly exciting—they are powerful tools for augmenting human expertise, not replacing it. Clinicians today have so many responsibilities, and AI is a promising way to help offset that work and allow them to focus more on patient care and activities that require a human touch.

          Explore AI in medicine

          Learn how artificial intelligence is helping physicians focus on patients, while technology supports efficient decision-making and clinical documentation with Nuance’s AI-powered solutions.

          Learn more

          Tags: , , ,