How artificial intelligence can make high quality care more affordable?

Quick Takes

  • AI can scale expert knowledge leading to improved patient access and quality of care as well as enhance patient experience and satisfaction, while balancing the rising costs of healthcare

  • Misconceptions about the purpose and intentions of AI can cause doubt and concerns over its use, but when carefully managed and made to work in unison with the creativity, innovation, and empathy of humans, it is a powerful tool

  • Successfully implementing AI requires an open mind-set and clearly defined business goals and processes that will benefit from a custom and dynamic AI system

Dr. Axel Nemetz

Head of Life Sciences, IBM Germany, Austria, Switzerland

Dr. Stefan Ravizza

Cognitive Business Decision Support Leader, IBM Switzerland

How artificial intelligence can make high quality care more affordable?

20 May 2019

Healthcare organizations today are undergoing transformation to improve a system that is no longer sustainable. Financial challenges – mainly due to rising costs and diminishing payments and reimbursements – and the shortage of skilled personnel are among the top pressures catalyzing this change. Faced with extensive pressure to do more with less, the implementation of artificial intelligence (AI), or more correctly “augmented intelligence”, is a way forward to reach sustainability. Yet, misconceptions and unrealistic expectations regarding the capabilities of AI hinder its potential. With the right partner and guidance on how to effectively implement AI to improve inefficiencies and challenges, the applications and benefits can be as diverse and creative as the visionary leaders that set out to use it.

The diverse applications of AI illustrate the power of the technology and its potential in the healthcare industry. Automating time-consuming administrative tasks are more obvious use cases of AI. However, it can also serve as a tool to support:

  • diagnostic decisions
  • remote patient services
  • development of novel diagnostics and therapeutics

Scaling expert knowledge to improve patient access and quality of care 

AI as a decision support tool for diagnostics means that medical expertise can be scaled up and patient access to high quality healthcare improved. As a result, AI can elevate precision medicine and positively impact the diagnosis and care pathways of important disease areas.

Individualized Patient Treatment

Watson Oncology Adviser at Memorial Sloan Kettering uses electronic medical records, real world evidence data, and the latest research and publications available to offer diagnostic and individualized patient treatment recommendations. In cases where diagnosis cannot be made accurately, it may even suggest missing diagnostic tests.1,2

The real value of the system, which is now used in over 200 hospitals worldwide, is that the supporting data for its recommendations are also transparently presented to the physician to review before making their clinical decision. Today, when medical knowledge is expanding exponentially (projected to double every 73 days by 2020)3, this is extremely useful as it is humanly impossible for healthcare professionals to keep up.

Liver Cancer

AI is also being used to assist in analyzing medical images to diagnose patients. Guerbet, a global specialist in contrast agents and solutions for diagnostic and medical imaging, are developing an AI software solution in collaboration with IBM Watson Health to support liver cancer diagnostics and care.4 Using computed tomography (CT) and magnetic resonance imaging (MRI) technology, AI will support clinicians via the automated detection, staging, tracking, monitoring, therapy prediction and therapy response of primary and secondary liver cancer.

Alzheimer’s Disease

A recent study showed that AI could also be used to detect subtle changes in glucose uptake by the brain in positron emission tomography (PET) scans. This allowed for the diagnosis of Alzheimer’s disease about 6 years earlier than the final clinical diagnosis.5

Stroke

Similarly, the use of AI to enable quick and expert level assessment of ischemic stroke from brain CT scans can serve as a valuable second opinion to help stroke patients receive timely life-saving treatment.6-8

Use cases such as these, show the potential of AI to augment the level of expertise of physicians and improve quality care access to patients, especially in areas where top specialists in a particular field are scarce or non-existent, or when time is critical.

Enhancing patient experience and satisfaction

With the increased use of mobile health apps, AI use can help increase patient satisfaction as well as decrease costs of healthcare organizations by aiding in the effective delivery of patient-centric remote services. For instance, a chatbot based on AI to diagnose patient symptoms in a collaboration effort between IBM and Medgate in Switzerland.9 The chatbot aims to optimize the healthcare system process from the initial patient visit – often times unsolicited – to the triaging of patients to the next appropriate step. This could range from a recommended teleconference with a disease specialist or suggested over-the-counter medicine. Patients benefit as unnecessary visits to the physician’s office or hospital are avoided, and they are put on the optimal, real world evidence based pathway of care right from the start.

As the population age increases, and patients expect more and more remote solutions to offer them the flexibility and convenience they desire, solutions supported by AI will become increasingly important.

Balancing the costs of research and development

Another sector of the health industry that can benefit from AI is drug and biomarker discovery. Within the current paradigm of personalized medicine (or precision medicine), there is a need to advance these areas. However, rising costs are unsustainable. Developing a new prescription medicine that gains market approval is estimated to cost $2.6 billion dollars.10 Estimates of developing and commercializing a new diagnostic biomarker can exceed $100 million.11

AI can be applied to optimize processes along the development pipeline, such as identification and validation of novel targets, repurposing of existing biomarkers or pharmaceuticals, aggregating and analyzing biomedicine information, and assisting in patient recruitment for clinical trials.12 These potential uses of AI provide the opportunity to counter the inefficiencies that arise in the classical development methods to lower costs and/or improve the standard of diagnostics and therapies available.

One such example is the development of a non-invasive blood test to detect early-stage Alzheimer’s disease.13,14 AI technology was used to identify a set of proteins in the blood that could predict the concentration of amyloid-beta – a biological marker associated with Alzheimer’s – in the spinal fluid. Thus, evading this expensive and invasive procedure. The idea is to enrich clinical trials with patients who are screened to have early-stage disease, as previous studies show potential therapies work best on these patients.15

Changing our beliefs – Augmented intelligence requires both humans and machines

Despite the many advantages, tension surrounds the use of AI. It is often criticized as a “black box”, in which the inputs and outputs can be viewed, without knowledge of how the machine works internally to the user.  As machine learning is based on the data used to train the system, concerns over bias built into the system also cause fear and doubt. The system can only be as good as the data used to train the system on, and human bias is a fact of humanity.  However, when this is understood, then precautionary measures can be taken to remove bias.

There are also misconceptions about the purpose of AI that are deeply rooted and exaggerated in science fiction.  AI is different from previous computer algorithms in which hard mathematical rules are set and the output is 100% accurate and true. AI can make mistakes and does not always come up with the “correct” answer 100% of the time.

The power of the technology, however, is to act with the computing power that humans don’t have the capacity for to make transparent judgments that are based on the evidence it is presented. This in combination with the creativity, innovation and empathy of human beings is where the real power lies. It is not to replace humans, but rather to augment our abilities to come to better or expert conclusions faster. When applied to the realm of healthcare, this translates to high quality care that is more affordable and accessible to everybody.

Technology is not the bottleneck, however changing cultural and behavioral beliefs about AI is the greater challenge. As the adoption of any technology – take the example of the use of Excel for accountants, or internet in schools – it takes time, yet the competitive advantage it may offer cannot be disregarded.

4 STEPS TO SUCCESSFULLY IMPROVE YOUR BUSINESS WITH AI

1.Have an open mindset and a clear understanding of the benefits for the business
2.Pick your battles wisely. With clearly defined goals of the processes that need to be redefined and made more efficient, it is possible to start transforming your business. Starting small and witnessing success quickly is important and will allow your organization to scale up and evolve from there, with a collective positive attitude towards the undergoing changes.
3.Co-create a customized system with a trusted partner to fit your needs. The solutions are not static, but will evolve along with the company, which is important to recognize and consider. Off-the-shelf systems do not work as well as every business and the workflows or inefficiencies that it sets out to solve are unique.
4.Keep the end-user at the center of the design process. It is essential that the end-user tests and gives feedback throughout the creation and optimization of the system.

We are at an exciting time for healthcare. There are challenges, but these challenges can be overcome. The technology exists to support our solutions, but it is up to the innovation and determination of leaders to take the first steps towards change.

Dr. Axel Nemetz is heading IBM’s life sciences business and helps companies to employ AI and other technologies to improve their business. He has spent more than 20 years working with pharma, medtech, diagnostics companies as well as with payors, providers and governments in Europe and the rest of the world.

Dr. Stefan Ravizza is leading the Swiss Cognitive Business Decision Support unit within IBM’s consulting unit. This spans the practices around artificial intelligence, data platforms, internet of things, and digital health. He has acted as a lead data scientist in many cutting-edge projects including in the pharmaceutical sector. He is the author of several patents and multiple publications including a recent publication in Nature Medicine.

References

  1. Memorial Sloan Kettering Cancer Center. (2014). Article available from https://www.mskcc.org/blog/msk-trains-ibm-watson-help-doctors-make-better-treatment-choices [Accessed April 2019]
  2. Bumrungrad International Hospital. (2015). Video available from https://www.youtube.com/watch?v=338CIHlVi7A [Accessed April 2019]
  3. Corish B. (2018). Article available from https://www.elsevier.com/connect/medical-knowledge-doubles-every-few-months-how-can-clinicians-keep-up [Accessed April 2019]
  4. Guerbet Global. (2018). Article available from http://www.guerbet.com/our-group/news/2018/news/article2/guerbet-and-2.html [Accessed April 2019]
  5. Ding Y et al. (2018). Radiology 290, 456-464
  6. Herweh C et al. (2016). Int J Stroke 11, 438-445
  7. Nagel S et al. (2017). Int J Stroke 12, 615–622
  8. Brainomix Ltd. Company webpage available from  https://brainomix.com/ [Accessed April 2019]
  9. CNN Money Switzerland. Video available from https://www.youtube.com/watch?v=P16d1ukZuXs [Accessed April 2019]
  10. DiMasi JA et al. (2016) J Health Econ 47, 20-33
  11. Dolginow D et al. (2013). Article availble from https://www.diaceutics.com/?expert-insight=mystery-solved-what-is-the-cost-to-develop-and-launch-a-diagnostic [Accessed April 2019]
  12. Mak KK and Pichika MR. (2019). Drug Discov Today 24, 773-780
  13. Goudey B. (2019). Article available from https://www.ibm.com/blogs/research/2019/03/machine-learning-alzheimers/ [Accessed April 2019]
  14. Goudey B et al. (2019) Sci Rep 9, 4163
  15. Godyń J et al. (2016). Pharmacol Rep 68, 127-38

Related articles

Information Technology

How to make the most of your healthcare data?

Karl-Heinz Fiebig

17 October 2019

7 ways digitization will shape the future of healthcare

Simone Edelmann

30 September 2019

Technology

Nanobots, humanoids & mixed reality: the future of healthcare?

Shivvy Jervis

2 September 2019

Technology

6 technologies that will transform healthcare

Simone Edelmann

5 July 2019