Author


Details

Likes
4
Followers
122
Directory
Healthcare

Tags





Sponsored

Re-imagining healthcare


  • Recently, Peter Arduini, CEO of GE Healthcare, proclaimed that the software development business “is central to our growth strategy
  • Although AI is in its infancy, AI technology has become embedded in all aspects of care journeys: from diagnosis to recuperation at home; from prevention to improved lifestyles
  • Notwithstanding, many established MedTech leaders still advocate the production of physical devices for episodic surgical interventions marketed by B2B business models in wealthy regions of the world
  • Jenson Huang, a key opinion leader from the AI industry recently stressed how rapidly AI technologies have advanced over the past decade and predicts that AI “will revolutionize all industries” over the next decade
  • If Huang is right and more MedTech leaders bet their future growth on innovative AI driven strategies, healthcare systems will be soon re-imagined

Re-imagining healthcare
 
On 16 February 2023, a Wall Street Journal article announced, GE Healthcare Makes Push into Artificial Intelligence”. The company, spun-out of General Electric (GE) in January 2023, is now an independent enterprise traded on Nasdaq, and Peter Arduini, its Chief Executive, says that the software development business “is central to our growth strategy”. In the first instance, GE Healthcare is planning to apply artificial intelligence (AI) and machine learning (ML) techniques to masses of disparate data generated by hospitals during patients’ therapeutic journeys, to enhance hospital services, improve patient outcomes and reduce healthcare costs.
 
Arduini is right. However, to fully appreciate the future potential impact of AI technologies on the medical technology industry and healthcare systems, we need to engage with key opinion leaders (KOL) from the AI industry. One such leader is Jenson Huang, a Taiwanese-American electrical engineer, founder, president and CEO of Nvidia, a semiconductor company launched in 1993. Today, it is a world leading, Nasdaq traded AI technology enterprise with a market cap of ~US$509bn, annual revenues of ~US$27bn and >26,000 employees. To put this into a perspective: if AI was the mid-19th century gold rush in the US, then Nvidia would be the producer of pickaxes for the hundreds of thousands of prospectors drawn to Sutter's Mill in Coloma, California. But before engaging with Huang, let us get a better understanding of the state of healthcare systems, AI and ML.
 
In this Commentary

This Commentary discusses Arduini’s proposition that AI big-data driven software strategies, which aim to enhance patient outcomes and reduce healthcare costs, are key to the growth of medical technology companies. This raises a question whether traditional MedTechs, producing physical devices, and marketing them with B2B business models will create sufficient growth and value over the next decade to satisfy their investors. Although AI technologies are in their infancy, they have already entered many areas of healthcare and are well positioned to play a significant role in future, re-imagined healthcare systems. The Commentary describes AI and ML, provides a brief history of AI, outlines its recent uptake in healthcare and notes how AI technologies have been used by both agile start-ups and giant techs to develop ‘big ideas’ with the potential to disrupt the medical technology market. We briefly describe six start-ups that have leveraged AI to enter the MedTech market and by doing so, increased the competitive pressure on traditional enterprises. Although AI technologies have only recently been introduced to healthcare systems, they are embraced by the FDA and feature in many aspects of patients’ therapeutic journeys: from diagnosis and treatment to recovery and rehabilitation at home. The Commentary takeaways suggest that the actions of industry leaders like Peter Arduini will have a significant impact of the shape on healthcare systems over the next decade.
 
Healthcare in crisis

Healthcare systems throughout the world are in crisis and experiencing large and rapidly growing care gaps,which we have described in previous Commentaries. These are created by growing shortages of health professionals and a vast and rapidly growing demand for care from aging populations; a significant proportion of which present with chronic lifetime diseases, such as heart disorders, diabetes, and cancer, that require frequent physician visits and more resources to treat. Such care gaps result in millions of people having difficulties gaining prompt access to health services, which delay diagnosis, worsen patient outcomes, and increase treatment costs. 

Addressing such issues requires re-imagining healthcare systems. Commercial enterprises have a role to play. Like GE Healthcare, agile start-ups and giant techs have embraced new and evolving AI technologies to create innovative offerings that provide solutions to care gaps predicated upon patient-centric, AI big-data strategies. However, many traditional medical technology companies have not developed software offerings and continue to focus on the production of physical devices, and B2B business models to support episodic hospital-based surgical interventions.  

 
Brief history of AI

AI refers to the development of computer systems that can perform tasks, which typically require human intelligence, such as decision making and natural language processing. The technology is based on the premise that machines can learn from data, identify patterns, and make recommendations with minimal human intervention. ML algorithms [instructions carried out in a specific order to perform a particular task] build mathematical models based on sample data, referred to as "training data", to make predictions or decisions without being explicitly programmed to do so.
 
AI has been around since the 1950s. The term was coined by computer scientist John McCarthy in 1956 at the Dartmouth Workshop in Hanover, New Hampshire, USA. In the early days of AI, scientists focused on building computers that could think, reason, and solve problems like humans. In the 1960s and 1970s, AI research concentrated on developing more advanced algorithms and techniques for programming computers to solve tasks. This resulted in expert systems, which used knowledge-based decision making to solve complex problems. In the 1980s, AI shifted towards ML, which allowed computers to learn from experience by enabling them to recognize patterns and make decisions based on data. In the 1990s, AI developed methods for robots to interact with their environment and learn from experience. This led to autonomous robots that can navigate and perform tasks in the real world. Today, AI research is focused on creating more intelligent and autonomous systems and is used in a wide range of applications, and increasingly in healthcare.
 
AI and healthcare

AI’s use in healthcare can be traced back to the 1970s, when researchers developed expert systems that could diagnose and treat certain medical conditions. Early AI healthcare applications were limited by the availability of data and the dearth of computer power. In the 1990s, as computing power increased and the internet became more widely available, AI began to be used more extensively in healthcare. One of the early applications was in radiology, where it was used to interpret medical images. Other applications included decision support systems for medical diagnoses and treatments, and natural language processing systems for medical documentation. In the 2000s, the use of AI continued to expand, with the development of ML algorithms that could analyze large datasets to identify patterns and make predictions. These were used in a variety of healthcare applications, including personalized medicine, drug discovery and medical imaging.
 
Today, AI benefits a wide range of healthcare applications from faster diagnosis to the prediction of pandemics, from clinical decision support to digital therapeutics. The aspiration of AI driven solutions and services in healthcare is super-human performance, free from errors and inconsistencies, and scalable to provide expert-level care across entire health systems. AI technologies have the potential to provide services that improve the accuracy and speed of medical diagnoses and treatments, monitor conditions, assist with recovery, support medicine regimens, facilitate personalized healthcare and reduce costs for providers. These functions are relevant in the context of attempts to narrow care gaps, but they require vast amounts of computing power, which most companies do not have in-house.
 
This is where cloud computing, and Nvidia's new solution come in. Dubbed "DGX Cloud", Nvidia’s offering is an AI supercomputer accessible via a web browser. The company has partnered with various cloud providers, including Microsoft, Google, and Oracle to develop the service, which provides enterprises easy access to the world’s most advanced AI platform and allows them to run large, demanding ML and deep learning workloads on graphic processing units (GPUs) to generate and implement ‘big ideas’.
 
Big ideas

New entrants to the medical technology market - agile start-ups and giant techs - often have ‘big ideas’; innovations with the potential to inspire stakeholders and disrupt the industry. By contrast, traditional MedTechs who do not employ AI strategies tend to have a dearth of big ideas and mainly focus their R&D spend on incremental improvements to their legacy devices. By contrast, new entrants have accelerated the use of AI, ML, and data analytics to help diagnose diseases earlier and monitor patients remotely. Further, they have championed wearable devices like Fitbits and Apple Watches that help people track their health metrics and allows them to make smarter decisions about their wellbeing. This is helping to transform the modality of healthcare from ‘diagnosis and treatment’ to ‘prevention and lifestyle’
 
Start-ups with big ideas
 
There are hundreds of healthcare start-ups with big ideas predicated upon innovative AI technology. To provide a flavour of these we briefly describe six.
 
Biofourmis
Boston based Biofourmis was founded in 2015. Its Biovitals™ Analytic Engine brings patient-specific data and ML together to provide the right care, to the right patients, at the right time. Advanced analytics process continuous and episodic data, notify clinicians of changes in patients’ conditions, and enable early intervention. With digital medicine, modular treatment algorithms (based on a patient’s disorder) enable the delivery of optimal medication.
 
You might also like: 
TytoCare
TytoCare is a New York-based medical technology start-up, founded in 2012, which aims to transform primary care by enabling people to have 24-7 medical examinations with a physician from the comfort of their home. The company has developed a suite of easy-to-use medical devices with built-in guidance technology and ML algorithms to ensure accuracy, which replicate face-to-face clinician visits. The devices include a hand-held modular tool for examining the lungs, throat, heart, skin, ears, and body temperature, and a health platform link to the cloud for storing, analysing, and sharing health data derived from the examinations.
 
Doctolib
Doctolib is a French digital health company founded in 2013. Its main product is a software-as-a-service platform for health professionals, which allows patients to book in-person and video consultations with healthcare providers. In January 2021, Doctolib acquired Tanker, a French start-up that developed the world’s first end-to-end encryption platform in the cloud, which Doctolib had been using since 2019. The Tanker platform is designed to be used by developers with no cryptographic skills and enables online businesses to easily encrypt their user’s sensitive data at the source: directly on end-users' devices. In October 2021, Doctolib acquired Dottori, an Italian online medical appointment scheduling service. The company is currently valued at >US$6bn, and  is used by ~300,000 healthcare professionals and ~70m patients in Europe.
 
CMR Surgical
CMR Surgical develops equipment and systems that aid in minimal access surgeries. During its establishment in 2014 in Cambridge, UK, the company’s founders asked, “why are so many people not receiving minimal access surgery and how can we change this?”. CMR’s main product is “Versius”, an EUMDR compliant device developed for high precision operations. During surgical procedures it can continuously collect data, which are stored and analysed to support surgeon training, and enhance the performance of future surgeries.
 
Healthy.io
Healthy.io is an Israeli start-up established in 2013. Its founders saw an opportunity to increase access to healthcare by leveraging the continuous improvement in smartphone cameras, which they transformed into at-home medical devices. As smartphone camera capabilities grew, Healthy.io’s range of clinical grade services expanded. With the company’s app and kits, users can undertake unitary tract infection (UTI) testing, prenatal monitoring, open wound assessments, and more, all in their homes. Health.io has partnered with healthcare systems throughout the world to provide clinical results at critical moments.
 
Proov
Proov, a US femtech start-up based in Boulder, Colorado, whose flagship offering is a rapid response progesterone test strip invented by Amy Beckley, a pharmacologist, with expertise in hormone signaling. It is the only FDA-cleared (March 2020) urine progesterone (PdG) test to confirm successful ovulation at home. Lack of, or insufficient ovulatory events, is the primary cause of infertility worldwide. In the US, ~12% of couples are diagnosed with infertility each year.  Thus, being able to confirm ovulation is an essential component of infertility evaluations in women.  Gold standards for confirming ovulation include transvaginal ultrasounds and serum progesterone blood draws. Both techniques are invasive, expensive, and/or inaccessible to most women. Proov’s offering is a non-invasive, inexpensive, home-based testing system.
 
A new era for AI in healthcare
 
Such start-ups with AI driven offerings suggest a new era for healthcare, which also is signalled in the introduction to a 2021, FDA action plan for AI/ML-based software medical devices. The plan describes how traditional B2B MedTech strategies are being complemented with B2C digital solutions and services that support entire patient journeys. According to the FDA’s action plan, “Artificial intelligence (AI) and machine learning (ML) technologies have the potential to transform healthcare by deriving new and important insights from the vast amount of data generated during the delivery of healthcare every day. Medical device manufacturers are using these technologies to innovate their products to better assist healthcare providers and improve patient care. One of the greatest benefits of AI/ML in software resides in its ability to learn from real-world use and experience, and its capability to improve its performance. FDA’s vision is that, with appropriately tailored total product lifecycle-based regulatory oversight, AI/ML-based Software as a Medical Device (SaMD) will deliver safe and effective software functionality that improves the quality of care that patients receive”. The agency currently has several ongoing projects designed to develop and update regulatory frameworks specific to AI. As of early 2023, there have been >500 FDA approved AI/ML-algorithms as medical devices.

 
Al and healthcare systems

Although AI is in its infancy and has only relatively recently begun to be used in healthcare systems, it has already taken root in many healthcare applications, including data analysis, diagnoses, monitoring, personalized apps, robotics, wearables, and virtual health assistance. This suggests a new era and the re-imagination of healthcare. Ambulances have become smart platforms, equipped with AI-based systems connected to hospitals, which can be used to diagnose medical conditions and provide real-time treatment recommendations. A&E departments use AI driven automated triage and diagnosis systems to assess incoming patients and prioritize those with the most serious conditions quickly and accurately. AI is also used to automate the dispensing of medications. Hospitals employ AI-based systems to analyze medical images such as X-rays and CT scans, which help medical personnel to quickly identify any abnormalities and make more accurate diagnoses. Surgery employs AI-enabled systems to assist with planning procedures, automating the delivery of anesthesia, and performing complex and delicate surgical interventions. Virtual recovery coaches use AI technology to create personalized plans for individuals recovering. Smart systems collect real time patient data and provide advice and support to help patients stay on track from their homes. AI-powered medication management systems help patients to track and manage their medications and send alerts to healthcare providers if there are any issues.
You might also like:

Have diversified medical technology companies blown their competitive advantage?
 
The cusp of a new era

According to Huang, a new era of AI has been triggered by a technology most people have become familiar with over the past few months: ChatGPT. Developed by OpenAI and built on top of its family of generative, large language processing models, which have been fine-tuned using both supervised and reinforcement learning techniques.
Huang views ChatGPT, as one of the greatest things that have been done in computing”. Generative AI models [algorithms that generate new outputs based on the data they have been trained on] have >100bn parameters and are the most advanced neural networks in today's world.  In no computing era has one computing platform (ChatCPT) reached ~150m people in ~60 days. In commercial terms this means, “a torrent of new companies and new applications . . . Nvidia is working with ~10,000 AI start-ups throughout the world in almost every industry”, says Huang. In a February 2023 earnings call to analysts Huang said that ChatGPT has incentivized businesses of all sizes to purchase Nvidia’s chips to develop ML software. Following the call, Nvidia’s market cap rose by US$79bn.
 
The democratization of AI programming

Huang’s enthusiasm for ChatGPT is partly because he perceives it as “democratising programming” by making human language a perfectly good computer programming language. The platform has the capacity to understand human-explained requests, generate coherent answers, translate texts, write code, and more. It has excited enterprises throughout the world and can be used for copywriting, translation, search, customer support, and other applications. While ChatGPT has many advantages, PyTorch and TensorFlow, two free and open-source software libraries have arguably done more to democratise programming by making it relatively easy to develop sophisticated ML applications without extensive programming skills. Notwithstanding, Huang is right to stress the significant leaps forward made by AI in the recent past and right to suggest that “AI is at a watershed moment for the world”.
 
Edge computing

Over the next decade, Huang predicts there will be a proliferation of edge-computing made possible by the spread of the Internet of Things (IoT). Edge computing is a connectivity paradigm that focusses on placing processing near to the source of data. This suggests that fewer activities will be executed using cloud computing. Instead tasks will be relocated to a user’s PC, cell phone or IoT devices. Huang refers to these as ‘AI factories’, which are positioned to have a significant impact on healthcare. By 2025, the global market for Internet of Medical Things (IoMT) is estimated to reach >US$500bn. This signals a significant change because currently most healthcare computing takes place in on-premises networks or, in the cloud. However, processing healthcare data from afar can be limited by infrastructures that cannot manage them quickly, securely, or cost-effectively. To address these issues, healthcare companies are implementing edge computing, which facilitates data being analysed and acted upon at the site of collection. This reduces end-to-end congestion and the constraints of limited connectivity and data broadband connections across vast distances by lowering transmission time, while also reducing risks to privacy and data protection. 

According to Huang, “AI processing performance has been boosted by a factor of no less than one million in the last 10 years”. Over the course of the next decade Huang predicts there will be, “new chips, new interconnections, new systems, new operating systems, new distributed computing algorithms and new AI algorithms (which will) accelerate AI by another million times."
 
Takeaways

Our discussion suggests that Peter Arduini, CEO of GE Healthcare, is right: software development is central to the growth potential of medical technology companies. Over the past two decades AI, ML and big-data strategies have substantially extended the horizons of industry players by giving them the means to provide software solutions and services to support entire patient journeys. This has introduced B2C MedTech business models, which complement conventional B2B models, and have the potential to provide access to new revenue streams while improving patient outcomes and reducing healthcare costs. If software initiatives like Arduini’s and others spread, healthcare systems are likely to be re-imagined. The fundamental technology of MedTech leaders is intelligence. But as Huang suggests, “We’re in the process of automating intelligence”, which can only empower industry executives. “The thing that’s really cool”, says Huang, “is that AI is software that writes itself, and it writes software that no humans can. It’s incredibly complex. And we can automate intelligence to operate at the speed of light, and because of computers, we can automate intelligence and scale it out globally instantaneously”. If Huang is right, over the next decade, AI is well positioned to play a significant role in re-imagining healthcare.

Comments