Impact of data analytics and artificial intelligence on the NHS
December 2017
1. What is the NHS?
In this report we will explore the impact data analytics could have in the healthcare industry, how can this affect and benefit the NHS and therefore all of its patients.
The NHS is the public universal healthcare system currently operating in England. It was launched in 1948 and employs more than 1.5 million people [1], putting it in the top five of the world’s largest workforces in the world.
The NHS is funded from taxation to the population. It manages a budget of £106.8 billion in 2016/17[2].
2. How can DATA ANALYTICS and AI disrupt the Healthcare industry?
In this report, we will concentrate on areas surrounding the patient from the moment that they are diagnosed to the moment they leave the hospital/ clinic. This is due to the fact that this is one of the most crucial parts in order to provide a better service and therefore maintain the health of the UK population.
In addition, it is also mentioned on their Principals and Values that “patients at the heart of everything it does” and that their objective is to provide “highest standards of excellence and professionalism” at “best value for taxpayers’ money and the most effective, fair and sustainable use of finite resources”
PwC reports that Healthcare services are the sector on which Data analytics is predicted to have the highest impact. It is also considered as the highest growing sector [4]. Mckinsey also outlined, that the EU public sector and healthcare have captured less than 30 per cent of the potential value we highlighted five years ago [5].
Healthcare Analytics Market is forecast to reach $29.84 billion by 2022 from $8.92 billion in 2017 [6] and $53.65 billion by 2025 [7].
In addition, compared to other developed or western countries, the penetration of digital solutions by hospitals to deliver healthcare services in the UK is quite low, falling the lowest in the chart with Australia.
The number of startups in the healthcare industry over the last 10 years has grown and in line with that, the amount invested into those companies is also growing.
Over 1.4k companies operate in this newly formed industry [8].
3. Why the use DATA ANALYTICS and AI could impact on the NHS?
The NHS has been facing a couple of problems in the last 10 years which have not only deceased their efficiency but also has created a bad PR with its customers. This is not ideal as the taxpayers of the UK are forced to fund this service.
As seen in the graphs above the population of the UK has been again steadily in the last 20 years and is predicted to keep growing at the same rate in the following 20 years [9].
As the population ages, a strain on the healthcare system arises. People over the age of 65 require more resources and attention than those younger, therefore, need a larger average spending per person
This means that the NHS needs a change in its operations to improve efficiency if it wants to be able to keep providing healthcare service to the UK mass market.
In this report, we will analyse how Data Analytics and AI could impact the NHS and in its operations.
The following areas of the Business Model Canvas (BMC) would be impacted by the introduction of Data Analytics:
Our conclusions for the report are that the implementation of such technologies could potentially impact the NHS by:
- Improve efficiency in hospitals
- Improve patient experience
- Help achieve 5 year forward view
4. How is DATA ANALYTICS and AI currently used in Hospitals?
In this section, we will explore two of the potential applications of Data Analytics and its possible effects on the NHS.
These were chosen due to the fact that we found with the use of Quid that this two were the biggest clusters with the largest number of applications.
4. 1 Intelligent Monitoring
Monitoring patients require (checking them and their health) regularly while they are in hospital and taking action if they show signs of becoming worse can help avoid serious problems [10]. This could occur following an emergency admission to hospital, after surgery and after leaving critical care.
Taking examples from some of the major hospitals in the USA, we can look at the application of AI in Cleveland Clinic’s to monitor their patients.
Cortana is integrated into the eHospital system, first launched in 2014. The system monitors “100 beds in six ICUs” from 7 pm to 7 am. It uses AI to predict at-risk patients by analysing the patient’s vitals and “prioritize activity for the benefit of all patients, and trigger interventions to accelerate patient flow” [11].
Data collected from monitored ICUs is stored in Microsoft’s Azure SQL Database. Collecting patient vitals and lab data and fed into the system to build a computer model integrating machine learning for predictive analysis.
Johns Hopkins uses a similar system as Cleveland Clinic’s and reports a 70% reduction in room holds, 30% faster bed assignments in the emergency department, and a 21% increase in patient discharges pre-noon [12].
Similarly, an EU project has also developed a system called Nightingale. The monitor will analyse the patient’s vitals during the night. Then if it deems a patient needs attention it will “assign an urgency level and select the right medic to alert” helping with the prioritisation of more than one emergency calls for doctors.
This is especially important as 75% of the time in a year is outside 9 am to 5 pm, Monday to Friday [13]. At most of these times, teams are made up of junior doctors with less experience. Therefore it’s more likely accidents could happen. This could help those doctors by giving them an ordered list of prioritization of their tasks instead of leaving it to their novice judgment.
It was found that junior doctors using this software completed their nonurgent tasks more rapidly (85.1 versus 157.6 minutes)[14].
Impact on the NHS:
- More efficient allocation of doctors and nurses time
- Less chance of committing mistakes and accidents occurring because of the wrong judgement of priority
- Help NHS follow through their 5-year plan by targeting “The funding and efficiency gap” and “The care and quality gap”
This would ultimately lead to better service with improved efficiency and knowledge about the patients current status. It will also have a direct impact on:
- Customer Relationship: this could potentially improve customer satisfaction and make customers rate the service higher as less further complications could occur but also mainly providing a better quality service at a faster rate.
- Key Resources: Intelligent monitoring software
- Key Partners: NHS might have to potentially partner up with tech or Data Analytics startups.
4.2 Intelligent Diagnosis
Misdiagnosis occurs approximately 5–15% of the time [15]. It has been found that 64% of doctors survey in the USA said that up to 10% of the misdiagnoses they have experienced directly resulted in harm to the patient. In addition, 28% of 538 reported diagnostic errors were life-threatening or resulted in the patient’s death or permanent disability.
But this is not only limited to the USA, the NHS paid out £ 194 million in compensation to 1,302 patients in 2014 [16]. On top of the financial cost, 12,500 NHS patients die a year due to blunders caused by staff [17].
The importance of giving the right diagnosis as fast as possible can completely change the future of the patient but also has a significant financial impact.
The startup Optellum is working to commercialize an AI system that diagnoses lung cancer by analyzing clumps of cells found in scans.
Doctors struggle to determine if a clump of cells is cancerous, so the diagnosis requires up to two-year follow-up monitor growth. This “increase patient anxiety, carry a risk of complications and present a huge and growing burden on healthcare system resources” [18].
Optellum uses “image-based risk stratification software” [19] powered by Deep Learning algorithms. As train data for their algorithms, they built the “world’s largest curated database of lung nodule patients” [20] specifically for machine learning.
Results suggest it could diagnose as many as 4,000 lung cancer patients per year earlier than doctors can [21]. Kadir (company’s chief science and technology officer) estimates that Optellum could cut costs by £10bn ($13.5 billion) if both the United States and Europe decided to utilize it [22].
But this type of technology is not only limited to Lung cancer. A group of scientists in Stanford have built an algorithm to diagnose skin cancer also using Deep Learning technologies. They used a training data of “130,000 images of skin lesions representing over 2,000 different diseases” [23]
To put this into context, in 2014, 2459 people died from melanoma skin cancer in the UK [24]. One person dies of melanoma every 54 minutes in the USA [25]. In addition, early detection correlates with a 97 per cent five-year survival rate in the USA.
Results show that the algorithm reports the same course of action (proceed with biopsy or treatment or reassure the patient) to the same ability as dermatologists [26].
Impact on the NHS:
- Better service towards patients which lead to higher customer satisfaction
- Less misdiagnosis, leading to lower costs
- Less misdiagnosis meaning higher survival rate and fewer complications for patients. This is in line with the NHS 5 year forward views “radical upgrade in prevention and public health”.
This would ultimately lead to better service with improved efficiency and knowledge about the patients current status. It will also have a direct impact on:
- Customer Relationship: this could potentially improve customer satisfaction and make customers rate the service higher as rates of survival could potentially increase and the percentage of complications and misdiagnosis would decrease.
- Key Resources: Cancer detecting Deep Learning Algorithms
- Key Partners: NHS might have to potentially partner up with University incubators where most of these tech startups are currently being formed.
5. Conclusion
Data Analytics and AI are currently shifting every industry including Healthcare and the operations inside Hospitals. The improvements such technology could bring to NHS are hard to measure but will likely impact every area of the service especially the relationship with the patient.
The key takeaways from this report are that Data Analytics and AI could:
Improve efficiency in hospitals:
- Significantly cut costs due to less misdiagnosis
- 30% faster bed assignments
- 70% reduction in room holds
- 50% faster completion of non-urgent tasks
Improve patient experience:
- Decreased chance misdiagnosis
- Decreased chance of complications
- Better judgment of priority
- Faster service at same or improved standards
Help achieve 5 year forward view :
- “Radical upgrade in prevention and public health”.
- Target “The funding and efficiency gap”
- Target “The care and quality gap”
All of the statements mentioned above would in return provide a possible improvement to face the current problems with the ageing population and the effect this is having on hospitals and that it will continue to have over the following decades.
APPENDIX
A. Data
The volume of Data of medical records, scans and diagnoses have been increasing on the last 30 years due to an improvement in the quality, knowledge and the wider access to healthcare.
The following graph breaks this down into different types and allows us to gauge how big this is by comparing it to data in companies like Facebook and Twitter:
B. Further detail on the technologies
Intelligent monitoring
Telehealth- earliest form of patient monitoring using technology.
Predictive Analytics
The diagram below illustrates how Microsoft’s Azure SQL Database works in order to deliver a prediction of the risk of the patient and prioritise the tasks of the doctors.
Intelligent Diagnosis
The way the diagnosis process occurs has changed significantly in the history of medicine. AI was first introduced in the ’60s and ’70s with the birth of computer science. Exact dates are difficult to find but during the 1990s and 2000s is when this technology started to be implemented in hospitals around the world.
Possible Obstacles and Limitations
In order to make this report more accurate, we should also consider the obstacles and limitations of the impact of such applications. The introductions of Ai and Data analytics into Hospitals and Healthcare may seem a very good choice on paper but the opinion of the general public has to be considered as they are the consumers of the service.
The introduction of Data Analytics would not be as simple as it may look. Some of the obstacles that NHS could potentially face is that, according to PwC, Germany and the UK are the only countries where unwillingness (51% and 50% respectively) is greater than willingness (41% and 39% respectively) to engage with AI and robots for healthcare needs.
Competitor Analysis
When choosing the NHS as the focus of the report, we faced the problem that the NHS has no real competitor. So in order to be able to compare we mostly use examples of major Healthcare Clinics and Hospitals in the USA throughout the report. In some cases, we also talk about other national healthcare systems in other countries.
This is not ideal as American hospitals and major clinics are private and therefore have different cultures towards efficiency and spending whereas the NHS is funded by taxpayers which shifts the actions and procedures of the organisation as they don’t focus on profit but in access of healthcare to the general public.
On the other hand, other National Healthcare systems are also not perfect competitor comparisons. Most of these are located in smaller countries and therefore have a different demographic consumer basis usually accommodating a rather smaller population.
C. Quid
When we first looked at the question our first two searches in Quid were the following, just to give me a general sense and idea of the industry and of the news stories related to the subject.
The second search is very similar but changed healthcare to NHS to test out what kind of reports and news stories were already out there.
The first search was then retired by eliminating the words financial services, consulting and insurance and this combined made up a large part of the companies targeting healthcare tech startups.
I ended up with the following 7 application clusters:
We can clearly see that diagnostics and monitoring are the two biggest clusters in the AI and Data Analytics in Healthcare.
We then delved deeper into the companies and saw that the number of startups has been increasing steadily through the last 10 years:
We can also point out that bigger investments are being made into such companies.
D. Applications
Before exploring our own applications and making a map, I started to look at other existing maps in order to better understand the industry before diving in. This also helped me identify a number of companies for my map.
After researching over 300 examples of startups and companies using AI and Data Analytics in healthcare and hospitals, we managed to find 44 applications of Intelligent Monitoring and 43 on Intelligent Diagnosis.
D. Bibliography
[1] About the National Health Service (NHS) — nhs.uk. Available at: www.nhs.uk/NHSEngland/thenhs/about/Pages/overview.aspx.
[2] Department of Health’s settlement at the Spending Review 2015. Available at: www.gov.uk/government/news/department-of-healths-settlement-at-the-spending-review-2015.
[3] Payments over £25k report: December 2017 — england.nhs.uk. Available at: www.england.nhs.uk/publication/payments-over-25k-reports-2017/.
[4] PwC — What doctor? Why AI and robotics will define New Health. Available at: https://www.pwc.com/gx/en/news-room/docs/what-doctor-why-ai-and-robotics-will-define-new-health.pdf
[5] The age of analytics: Competing in a data-driven world — Mckinsey. Available at: https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/the-age-of-analytics-competing-in-a-data-driven-world
[6] Healthcare Analytics/Medical Analytics Market — ReportsnReports, prnewswire.com. Available at: www.prnewswire.com/news-releases/healthcare-analyticsmedical-analytics-market-273-cagr-to-2022-led-by-north-america---reportsnreports-660521093.html
[7] Healthcare Analytics/Medical Analytics Market — Grand View Research, Inc. Available at: https://www.grandviewresearch.com/research-insights/healthcare-predictive-analytics-market-insights
[8] Quid Platform. Available at: https://quid.com/
[9] Overview of the UK population: July 2017 — Office for National Statistics. Available at: https://www.ons.gov.uk/releases/overviewoftheukpopulationjuly2017
[10] National Institute for Health and Care Excellence. Available at: https://www.nice.org.uk/
[11] Big Data, Analytics & Artificial Intelligence The Future of Healthcare is Here- General Electric. Available at: https://www.gehealthcare.com/static/pulse/uploads/2016/12/GE-Healthcare-White-Paper_FINAL.pdf
[12] Patient Safety Improved with Centralized Hospital Command — Relias Media. Available at:https://www.reliasmedia.com/articles/139995-patient-safety-improved-with-centralized-hospital-command
[13] Larkin, Chris, et al. “‘Night Shift’: A Task Simulation to Improve On-Call Prioritisation, Self-Management, Communication, and Route Planning Skills.” 2014
[14] Larkin, Chris, et al. (2014)“‘Night Shift’: A Task Simulation to Improve On-Call Prioritisation, Self-Management, Communication, and Route Planning Skills.” 2014
[15] Berner, Eta S., and Mark L. Graber. (2008) “Overconfidence as a cause of diagnostic error in medicine.” The American journal of medicine 121.5: S2-S23.
[16] Blunders by doctors cost NHS £4million a week — DailyMail.co.uk. Available at: www.dailymail.co.uk/news/article-3206686/Blunders-doctors-cost-NHS-4million-week-Health-service-paying-millions-compensate-patients-misdiagnosed-doctors.html#ixzz5D8jBh0mk.
[17] Blunders by doctors cost NHS £4million a week — DailyMail.co.uk. Available at: www.dailymail.co.uk/news/article-3206686/Blunders-doctors-cost-NHS-4million-week-Health-service-paying-millions-compensate-patients-misdiagnosed-doctors.html#ixzz5D8jBh0mk.
[18] Mirada Medical and Optellum announce collaboration to accelerate Deep Learning based lung cancer diagnosis products and to present scientific results at the RSNA 2017 — Miranda Medical. Available at: mirada-medical.com/mirada-medical-optellum-announce-collaboration-accelerate-deep-learning-based-lung-cancer-diagnosis-products-present-scientific-results-rsna-2017/.
[19] Mirada Medical and Optellum announce collaboration to accelerate Deep Learning based lung cancer diagnosis products and to present scientific results at the RSNA 2017 — Miranda Medical. Available at: mirada-medical.com/mirada-medical-optellum-announce-collaboration-accelerate-deep-learning-based-lung-cancer-diagnosis-products-present-scientific-results-rsna-2017/.
[20] Mirada Medical and Optellum announce collaboration to accelerate Deep Learning based lung cancer diagnosis products and to present scientific results at the RSNA 2017 — Miranda Medical. Available at: www.mirada-medical.com/mirada-medical-optellum-announce-collaboration-accelerate-deep-learning-based-lung-cancer-diagnosis-products-present-scientific-results-rsna-2017/.
[21] AI early diagnosis could save heart and cancer patients — BBC News. Available at: www.bbc.com/news/health-42357257
[22] AI early diagnosis could save heart and cancer patients — BBC News. Available at: www.bbc.com/news/health-42357257
[23] Deep learning algorithm does as well as dermatologists in identifying skin cancer — Stanford News. Available at: https://news.stanford.edu/2017/01/25/artificial-intelligence-used-identify-skin-cancer/
[24] Skin cancer statistics — Cancer Research UK. Available at: https://www.cancerresearchuk.org/health-professional/cancer-statistics/statistics-by-cancer-type/melanoma-skin-cancer/incidence
[25] Cancer Facts and Figures 2018. American Cancer Society.. Available at: www.cancer.org/cancer/melanoma-skin-cancer/about/key-statistics.html
[26] Deep learning algorithm does as well as dermatologists in identifying skin cancer — Stanford. Available at: news.stanford.edu/2017/01/25/artificial-intelligence-used-identify-skin-cancer/
