How artificial intelligence in diagnostics is transforming healthcare
How artificial intelligence in diagnostics is transforming healthcare11 October 2021 | 6min
Artificial intelligence in diagnostics has the potential to make healthcare more accessible, affordable, and efficient
3 ways AI can potentially have a strong impact include enhancing efficiency and accuracy of diagnostics, improving image recognition, and alleviating administrative and laboratory resource pressures
Experts agree that trust and transparency are essential for AI-based tools to reach their full potential
Artificial intelligence (AI) has powerful potential within healthcare, promising the ability to analyze vast amounts of data quickly and in detail. In a field such as in vitro diagnostics (IVD), this could have transformative implications.
First, AI in diagnostics has the potential to make high quality healthcare more accessible and affordable by assisting healthcare providers to more quickly make the most appropriate treatment decisions for their patients. Second, AI could potentially transform the back office by performing otherwise tedious and time consuming administrative tasks, enabling staff to focus on those that add value, while decreasing inefficiencies and improving the use of resources.
Let’s take a closer look at how AI is revolutionizing diagnostic and care pathways.
AI in diagnostics to enhance efficiency and accuracy of clinical decisions
As we move towards a digital world, the global datasphere is projected to grow from 33 zettabytes in 2018 to 175 zettabytes by 20251 (equal to 175 billion terabytes). Compared to manufacturing, financial, and media & entertainment industries, healthcare data is set to grow the most with a compound annual growth rate of 36%.1
Given this vast amount of data being generated (including the advancement of medical knowledge), physicians today are faced with an overwhelming amount of information when working to diagnose even a single patient. AI, however, has the potential to provide healthcare professionals the ability to speed up and improve their diagnostic capabilities by helping to extract clinically-relevant insights from the wealth of information available.
The power of AI to help diagnose diseases at their early stage was recently highlighted in a study that evaluated its use to identify COVID-19 positive patients. An AI algorithm that integrated chest computed tomography (CT) findings with clinical symptoms, exposure history and laboratory testing was shown to perform equally well in correctly identifying COVID-19 patients compared to a senior thoracic radiologist.2 Additionally, the algorithm outperformed radiologists in identifying patients positive for COVID-19 via reverse transcription-polymerase chain reaction (RT-PCR) who presented with normal CT results in the early stage of disease.2
This study showed the potential of AI as a useful screening tool to aid in the quick diagnosis of infectious diseases such as COVID-19, especially when access to specialists is limited and time is of essence in order to start appropriate treatment.
AI in improving image recognition in the diagnostic work-up
AI technologies are making great strides in medical imaging. Studies have shown that the use of AI may be able to enable earlier disease detection, while also enhancing the workflows by accelerating reading time and automatically prioritizing urgent cases.2,3
AI can look at vast numbers of medical images and then quickly and regularly identify patterns, including variations that humans cannot.3 This may not only improve patient outcomes, but also save money – earlier diagnosis and treatment of many cancers, for example, may cut treatment costs by more than 50%.4
There is applicability for AI systems to aid in making diagnoses based on medical imaging across many disease areas including oncology and cardiology, gastroenterology or hepatology, and neurology.5 This adds to the huge potential of AI to support clinical decisions in time critical situations or when there is a lack of expert knowledge available, such as in remote or poorly funded medical facilities.
Alleviating administrative and laboratory resource pressures with AI
From 1990 to 2012, the U.S. healthcare workforce grew by 75 percent – but 95% of that growth was in administrative roles, not physicians.6 There are now 10 administrators for every doctor.6 This rapid growth in administration is driven by increasingly complex regulations, technology and inefficiencies in the system.6
A slowdown in the use of technology or a relaxation of regulatory requirements are unlikely, so harnessing tools like AI-powered back office software to help offset this holds incredible value. A recent report by Accenture estimates that by 2026, AI applications that streamline the administrative workflow of healthcare organizations could lead to annual cost savings of US$18 billion in the US alone.7
AI also has the potential to help address the shortage of laboratory staff/technicians that has plagued the field for years. The US Bureau of Labor Statistics estimates about 25,900 openings for clinical laboratory technologists and technicians each year, on average, between 2020 and 2030.8 However, less than 5000 students are graduating from accredited programs each year.9
While AI has incredible potential for analysis and diagnosis as discussed above, the majority of time and effort in a lab is spent on pre- and postanalytical processes.10 AI could help bring significant improvements to the workflow and operations, saving time, labor, and costs.10
AI experts provide insights into its successful implementation
To be successful, the use of AI in diagnostics needs to be guided. Machines are very good at specific but limited tasks – narrow and deep. Humans will need to play a critical role in defining the intended use of any AI-based tool, as well as its design and implementation.
We have spoken to several experts in recent years, and here are some of the perspectives and insights they have contributed to the discussion around AI in healthcare.
Vivienne Ming, Founder and Executive Chair of Socos Labs
AI can do a superhuman job at a fraction of the cost than a human being. Yet, there are flaws and concerns about AI. Vivienne Ming dives into why defining the right question and knowing how to solve it must be done on the human level before expecting AI to get it right for us. She also shares her insights and experience about building AI that is both ethical, and powerfully impactful.
Carolyn Lam and James Hare, founders of Us2.ai
Carolyn Lam and James Hare share their advice to other companies wishing to transform traditional practice or methods: Ignore being trendy. Focus on providing value to your customers and their patients, believe passionately in your mission and work incessantly. The rest will work itself out.
Dr. Axel Nemetz and Dr. Stefan Ravizza from IBM
Axel Nemetz and Stefan Ravizza shared their 4 steps to successfully improve your healthcare business with AI: have an open mindset, pick your battles wisely, co-create a customized system with a trusted partner and keep the end-user at the center of the design process.
Karl-Heinz Fiebig, cofounder and chief innovation officer of idatase GmbH
Hospitals and commercial labs have a lot of patient data, but it may be unstructured or fragmented. Karl-Heinz shares his top 3 tips companies can start doing now to make the most of their healthcare data with AI: allow easy access to data, structure data, and ensure data is easily machine readable.
Nell Watson, tech ethicist, and Jim Stolze, entrepreneur
Nell Watson and Jim Stolze discuss IT security and privacy, and how these will be an enabler of AI, and must be built into any AI diagnostic system.
The future success of AI in diagnostics relies on trust
As the use of AI in diagnostics steps out of its infancy and continues to progress, the opportunities it presents to improve healthcare and make more efficient use of resources seems limitless. Of course, as with any disruptive technology, there are remaining questions and concerns around its use.
In our discussions, experts largely stress the importance of trust and transparency in establishing the integrated use of AI-based solutions in healthcare systems. What is also clear, is that developing such tools will require collaboration across many disciplines with the patient at the center and the end-user as an integral part of the design.
Simone Edelmann, PhD is an editor and contributor at HealthcareTransformers.com. After completing her PhD from the Institute of Biotechnology at the University of Lausanne, Switzerland, she found her passion in medical and scientific communications. She is dedicated to delivering high-quality content on the topic of the future of healthcare to our readers.
- Reinsel et al. (2018). Article available from https://resources.moredirect.com/white-papers/idc-report-the-digitization-of-the-world-from-edge-to-core [Accessed October 2021]
- Mei. (2020). Nature Medicine 26, 1224–1228
- Pesapane. (2018). Eur Radiol Exp 2,35
- Aboshiha et al. (2019). Article available from https://www.bcg.com/en-ch/publications/2019/chasing-value-as-ai-transforms-health-care [Accessed October 2021]
- Liu et al. (2019). The Lancet Digital Health 1, E271-E297
- Ross. (2019). HealthLine News. Article available from https://www.healthline.com/health-news/policy-ten-administrators-for-every-one-us-doctor-092813 [Accessed October 2021]
- Accenture Health. (2017). Report available from https://www.accenture.com/_acnmedia/PDF-49/Accenture-Health-Artificial-Intelligence.pdf [Accessed October 2021]
- US Bureau of Labor Statistics (2021). Report available from https://www.bls.gov/ooh/healthcare/clinical-laboratory-technologists-and-technicians.htm [Accessed October 2021]
- Data USA. (2021). Report available from https://datausa.io/profile/cip/clinical-laboratory-technician [Accessed October 2021]
- Halasey. (2019). Article available from https://clpmag.com/diagnostic-technologies/anatomic-pathology/microscopy/ais-impact-on-in-vitro-diagnostics/ [Accessed October 2021]