Beyond Unimodal Data: Embracing Multimodal AI

Med-PaLM M represents the first demonstration of a generalist multimodal biomedical AI system, transcending traditional unimodal data analysis. The system possesses the capability to encode and interpret diverse types of medical data spanning text, images, genomics, and more within the same model architecture. This approach harnesses the power of multimodal data, enabling the model to gain a comprehensive and nuanced understanding of a patient’s health status.

MultiMedBench: The Fuel Powering Med-PaLM M

Key to the development and evaluation of Med-PaLM M is MultiMedBench, a curated multimodal medical dataset. Encompassing over 1 million examples for question answering, report generation, classification, and other clinically relevant tasks, MultiMedBench provides a comprehensive benchmark for training AI systems in diverse biomedical applications. This impressive resource is a testament to the significant progress in biomedical data generation and innovation.

Setting a New Benchmark in Performance

Med-PaLM M consistently reached or exceeded state-of-the-art performance across all tasks in the MultiMedBench benchmark. This represents a seismic shift in biomedical AI, illustrating the potential of a single, flexible model to compete with or surpass specialized models optimized for individual tasks.

Revolutionizing Patient Care with Med-PaLM M

With its ability to integrate multimodal patient information and facilitate the positive transfer of knowledge across medical tasks, Med-PaLM M has significant potential to enhance diagnostic and predictive accuracy. Preliminary evidence also suggests that Med-PaLM M can generalize to novel medical tasks and concepts, performing zero-shot multimodal reasoning.

For instance, the model has demonstrated the ability to accurately identify and describe tuberculosis in chest x-rays without any prior exposure to such cases. These emergent capabilities not only underline the versatility and adaptability of the system but also hint at its potential utility in data-scarce biomedical applications.

Med-PaLM M: A Partner in Care

A clinical evaluation of Med-PaLM M’s reports revealed a clinically significant error rate on par with human radiologists, emphasizing the system’s potential for real-world clinical application. This, coupled with the model’s analytical speed, allows clinicians to make swift and accurate decisions, enhancing patient care outcomes.

While much work remains to be done in validating these models in real-world scenarios, the results achieved by Med-PaLM M represent a milestone in the development of generalist biomedical AI systems. The promise of such models is immense, ranging from biomedical discovery to care delivery.

As we stand at the precipice of a new era in AI and healthcare, the unveiling of Med-PaLM M underscores the power of artificial intelligence as an indispensable partner in care for clinicians. It’s not just an option, but an imperative – a promising stride in the continuing journey to improve patient outcomes and revolutionize healthcare. The future of AI in healthcare is here, and it’s called Med-PaLM M.

Top of Form

Bottom of Form

The post A New Era in Medical AI: The Power Of Med-PaLM M appeared first on Medika Life.

Can ChatGPT and OpenAI compete with human creativity? Its ability to process is rooted in machine learning — the ability to scour in a blink a wealth of databases, and pirated content, and mimic the human experience. Can a computerized clone of humanity’s documented contributions be as vibrant and present as a human? Right now, my experience as a user is absolutely not! Technology cannot differentiate between information and its dark side, misinformation. It lacks head, heart and gut at this time. That will change in time.

Poets and creative writers and strategists and scientists fear not. Curious, creative and inquisitive minds will master this technology rather than let the technology sideline humanity. While the typewriter was always nearby, Nobel Prize author Ernst Hemingway wrote — and rewrote — many of his famed novels by hand. Yet, he didn’t shun the technology of his day. He found that typing what he had written by hand allowed him to improve it.

Bicycles and Horseless Carriages

Yes, bicycles and horseless carriages led to the horse and buggy’s retirement. That’s one machine’s efficiency versus the machine’s efficacy. But the person or health system that embeds AI into its patient care model will outrun those that shut their doors to keep it out.

Likely the inquisitive radiologist who adds AI into the mix will arrive at a diagnosis faster and with greater accuracy. They won’t abandon human judgment; they will use technology to go beyond the general report rising to higher and higher levels of diagnostic guidance.

Beyond enhanced predictive ability, AI will elevate gifted and early adapter medical practitioners to a heightened level of performance. While leading doctors may not always have the greatest bedside manners, they demonstrate consistently heightened clinical ability. AI must never be a medical replacement for learned minds and human connections — it augments their well-honed clinical skills.

“I Don’t Have Time to Read EMRs”

During the last 12 days of my noted physician father-in-law’s life, a nurse at a leading hospital replied to my questions about his care with a worrying admission: “I don’t have time to read patient records.” That admission is symptomatic of systemic failures. EMRs are not standardized; patient data cannot easily be shared between providers or accessed by patients visiting multiple hospitals. This leads to duplicative tests and procedures and can lead to medical errors and misdiagnoses, sometimes with catastrophic consequences. And it also leads to health professional burnout!

In FRAGMENTED: A Doctor’s Quest to Piece Together American Health Care physician-author Ilana Yurkiewicz, MD, outlines how cracks in the healthcare system cost us time, energy, and lives. With a combination of real-world experiences, she strikes at the heart of the assumption when you see a doctor: the doctor knows your medical story and makes decisions based on that story. Wrong.

We are our medical historians. When we switch doctors, we begin again to create our medical history. Critical details of life-saving treatment plans need to be made aware of muddled electronic charts. The doctors we see change according to specialty, hospital shifts, or an insurer’s whims.

Tom Lawry, the bestselling author of Hacking Healthcare, the former national director of AI for Microsoft, and now a global advisor on how AI impacts health systems, comments on a UPenn Health study using AI analyzed all inpatient and outpatient notes written in the EMR from 2015 to 2020. Analyzing more than 1.96 million unique patient entries, 50.1% of input was duplicated from prior notes about the same patient — no wonder the nurse didn’t have time to read my father-in-law’s EMR report. It’s not surprising that Lawry believes “AI and the intelligence revolution will reboot an ailing system.”

We Have Nothing to Fear But Misaligned IQ and EQ

Most worries about AI are rooted in our imagination — not our experiences. Advances in artificial/augmented intelligence (AI) have the potential to advance patient care and improve healthcare outcomes. ChatGPT, as an AI-powered conversational agent, offers numerous advantages in delivering care. It thrives on heightened creativity and cognitive effort. It needs humanity to be additive.

Here is my Top Seven List of expected advantages AI and ChatGPT will deliver to patient care:

1. Enhanced Patient Engagement and Education: AI-powered conversational agents like Chatbot can engage patients in conversations, collecting and synthesizing information about their condition, treatment options, and self-care practice. This interactive engagement can empower patients to take an active role in their health — most notably by speaking openly about their needs leading to better adherence to treatment plans and improved health outcomes.

2. 24/7 Access to Healthcare: ChatGPT offers round-the-clock accessibility, allowing patients to seek healthcare advice anytime. This feature is particularly beneficial for patients in rural areas, those with limited mobility, or those needing immediate medical assistance — a medical AI researcher. AI-connected systems can quickly analyze, triage, and move patients into the care system.

3. Predictive Analytics for Disease Detection: AI algorithms can analyze patient data, including medical records, sensor data from wearables, and lifestyle factors, to identify patterns and risk factors associated with various diseases. Leveraging predictive analytics, AI can detect early warning signs and predict the likelihood of disease (re)occurrence. This enables health providers to intervene, implement preventive measures, and initiate early treatment, improving patient outcomes and reducing healthcare costs.

4. Efficient Triage and Decision Support: ChatGPT can assist in the triage process by asking questions and gathering patient information, helping to identify urgent cases, and redirecting them to the right health professionals. AI algorithms can analyze patient data and provide decision support to healthcare providers, aiding in accurate diagnoses and treatment recommendations. Ideally, the technology can do a better job helping medical staff to look at the “whole patient” and connect the dots — essential for “Zebra-like patients.”

5. Personalized Care and Remote Monitoring: AI algorithms can analyze patient data, including medical records, lab results, and wearable device data, to generate personalized insights. ChatGPT can use the information to provide tailored recommendations for lifestyle modifications, medication adherence, and disease management. Additionally, AI-powered monitoring systems can track patient vitals remotely and alert healthcare providers in case of abnormal readings as we shift more and more to home recovery and care.

6. Medical Imaging Analysis: AI algorithms can analyze medical images, such as X-rays, MRIs, and CT scans, to aid radiologists and clinical experts in the detection and diagnosis of conditions. These algorithms can identify abnormalities and predict disease progression. AI-based image analysis amplifies the professional’s efficiency, reducing human error, and speeding up the diagnosis process. The net result will be better patient outcomes.

7. Data-driven Health Research: AI algorithms can analyze vast amounts of patient data to identify patterns, detect trends, and generate insights that inform healthcare research and policy-making. This was key to addressing COVID-19 and accelerating vaccine-like treatments to mitigate illness. By leveraging AI and ChatGPT, researchers gain deeper insight into disease progression, treatment effectiveness, and population health trends. These insights can drive evidence-based decision-making and contribute to advancing medical knowledge. This will be critical to rallying to resolve the next pandemic.

Is it Really Human Versus Machine?

AI and ChatGPT almost guarantee several potential advantages in patient care. It’s not a perfect system nor a replacement for people. John Nosta, a global innovation theorist in speaking with Artur Olesch, founder and editor-in-chief of aboutDigitalHealth.com, said:

“While AI has shown promise in areas such as image recognition, natural language processing, and autonomous systems, there are still areas where AI struggles, such as understanding context, making subjective judgments, and dealing with incomplete or noisy data. Additionally, there are ethical considerations around the use of AI, including issues related to privacy, bias, and fairness.”

I still prefer to rely on Microsoft Spell Check than memorizing Webster’s Dictionary. I also prefer a physician who deploys human skill, amplified knowledge and the might of another 10,000 physicians’ aggregated experience to analyze and treat serious disease. In an AI-influenced world, we will still need to secure patent protection and copywrite guarantees for unique intellectual property and creative effort. But closing the door on this technology is to condemn too many people to an ill-fated end.

The post Will AI Replace Humanity? What Happens if it Makes Us More Human When It Comes to Health? appeared first on Medika Life.

A Duke University professor of bioscience, Dr. Nita Farahany, has published a book on the subject, and what it reveals may be intellectually exciting, but also seriously concerning. The broad scope of the book’s topic I’ve summarized below, along with some additional information.

Two distinct businesses, Meta and Neuralink, are investigating the potential for using innovative technology to reveal people’s thoughts. The ultimate aim of both businesses—which are addressing it in different ways—is to create tools that can record and decipher the neural activity of the brain.

Meta, formerly known as Facebook, is looking into the development of brain-computer interfaces (BCIs) that allow for the direct recording of neural activity from the brain. Scientists want to create a wearable gadget that can record electrical signals produced by brain activity and convert them into digital signals that a computer can understand.

Users will talk to digital systems with their minds thanks to the device, which works like a keyboard or mouse. Even though we don’t know all of the details about their technology yet, it is thought that their device would accurately read brain activity by using electrodes and machine learning algorithms.

On the other hand, Neuralink is investigating the creation of a brain implant that can communicate with the brain directly. The implant consists of a tiny chip that is placed into the skull and coupled to electrodes that are inserted into the brain. It is currently being tested on animals. The ability of the electrodes to pick up neural activity lets the brain and digital systems talk to each other in both directions and makes certain parts of the brain work.

Neuralink’s goal is to create a tool that can treat neurological conditions like Parkinson’s and Alzheimer’s, but the technology can also divulge private thoughts. The idea of the “thought police” is becoming a reality. It has shades of the book “1984” only this one is a definite reality.

Though they are both still in the early phases of research, Meta and Neuralink’s technologies have the power to completely change the way we interact with digital systems and disclose private thoughts. But this technology also brings up important ethical and legal questions about how these devices can be used and abused.

One of the biggest worries about brain-computer interfaces is that they could spy on people. Devices might be able to reveal private information like opinions, thoughts, and emotions if they can accurately read cerebral activity. They might employ this information in criminal investigations or even to influence people. As a result, there might be a need for legal action to control the use of such devices and safeguard people’s privacy.

The potential for such technology to be abused is another worry. For instance, there would be a risk of cyber-attacks or hacking if a brain-computer interface were used to control a digital system like a car or an airplane, which may have disastrous effects. Also, this technology could make mind-controlled soldiers or other mind-controlled laborers, which would take away people’s freedom and be against human rights. Are we headed for the world of Total Recall?

The development of brain-computer interfaces has the potential to completely change how we use digital systems and reveal our deepest thoughts. However, it also raises important ethical and legal questions. It is crucial to carefully weigh the possible advantages and risks of such technology and to make sure that moral and legal frameworks are in place to safeguard people’s liberty and privacy.

Obviously, one of the problems with this technology is the lack of understanding in legal circles about it. If people in Congress can’t follow explanations of current technology, how will they respond to highly complex systems?

The post “Cognitive Liberty” May Be Imperiled in a World of AI Intrusion appeared first on Medika Life.

The digital health sector has grown significantly during the past decade, primarily mobilized to address unmet health-system and care needs. From $2 billion invested in 2011, that figure morphed 28-fold to $56 billion in 2021. In the shadow of the COVID pandemic, unprecedented investment poured into digital health during this two-year window, accelerating the adoption of AI, remote patient monitoring, telehealth, digital therapeutics, wearables, VR and more.

DOWNLOAD THE FULL REPORT HERE

Galen Growth and FINN Partners have collaborated – for a second time – to offer the entrepreneurial invention and investment industry a wealth of data that can guide financing life-sustain – perhaps saving – enterprises.  With this detailed report, data is shared not as “one-off” ideas but as puzzle pieces assembled to provide actional insight into a dynamic and evolving digital innovation sector that offers the transformative power to find effective detours around health ecosystem fragmentation.

The Damocles Sword of Access and Outcomes

The two companies avoided hype and biases to create this analysis enabling investors or potential business partners to weigh subjective decisions alongside objective facts in recognizing that the fragmented health ecosystem centers around preserving the system itself are vital to helping product innovators and patients navigate the healthcare labyrinth. That economic return on investment is weighted along with patient experience and outcomes.  This is the Damocles sword that inventors and investors must always consider – reimbursement that ensures access to care and meaningful and measurable outcomes.

While digital health venture capital fell short of the exorbitant funding of 2021, it closed ahead of 2020 levels. A closer look at the year-on-year development of funding value by funding stages shows that funding in 2022 nearly maintained the level of 2021 for the Early Stage and Series A funding rounds. However, it lagged behind the 2021 values across late-stage enterprise funding.

Investors learned and began to put their due diligence into action – targeting their energies toward specific geographies, sectors and needs (e.g., clinical trials).  They looked for therapeutic categories receptive to digital health innovation.  Within the disease-specific investments, oncology saw the highest participation of investors (9%), followed by mental health (7%) and women’s health (7%).  C-Suite leaders and their funders will do well to study the report data and to determine next steps.

Just Facts – 250+ Million Data Points Analyzed

While other studies highlight “media grabbers” and numbers disconnected from insight into where the money is going — from the investment stage to disease states— The Global State of Digital Health connected more than 250 million critical analytic dots. This report will guide C-Suite leaders, business development and therapeutic area heads, private equity funds, and financial analysts to make savvy decisions that drive return on investment—whether the goals be financial or patient care.

After an extraordinary two-year pandemic funding spree, there are apparent shifts in how investors place their economic bets. In this year of geopolitical turmoil and uncertain financial performance, digital health remains vital to the future of the health economy and in addressing public health press points. However, the development stages in which funds are directed, therapeutic categories selected, provider systems that demand transformation, and the regions where innovators establish headquarters are now investment factors.

Bottomline, funding levels exceed the pre-pandemic values. Digital health ventures are busy – submitting more than double the number of regulatory filings in 2021 than before the COVID era. There is a dramatic increase in partnering activity over 2021. If dollars and Euros are the only indicators of energy and commitment, it is easy to see why hand-wringing news stories prevail. Look at all the metrics that determine a vibrant sector.

In combining investment, regulatory filings for new product introductions, and partnerships, it is apparent that 2022 has been a formative period. Investors and innovators are putting in place the processes and systems – the people and science – that drive adoption and market traction. Investors are no longer looking at the sector as monolithic.

Putting the Right People and Processes on the Innovation Bus

If 2021 was the year of exuberance, 2022 was, as Jim Collins, author of the classic business bestseller “Good to Great,” dedicated to putting the right people – leaders and digital health sectors – on the bus. For example, medical diagnostics ventures had an exceptional year by representing 12 percent of regulatory filings of Digital Health ventures across the globe, taking the top spot for ventures with significant Clinical Evidence (Evidence Signal >40), and capturing the highest investment volume share (17%), with the total cluster funding value falling short of 2021 by 16 percent.

As the report suggests, 2022 is a turning point for digital health – a shift from Digital Health 1.0 to the 2.0 era, with Digital Health evolving into a more mature ecosystem. The recent data reaffirm that perspective and a fast-paced realignment occurring within the sector.

Looking at the past year, it is evident that investor mindsets matured in 2022, and lessons learned were applied. Obstacles – from regulatory hurdles to payer pushback – made incubators, accelerators and private equity groups much more selective in their investment and partnership decisions. They looked beyond “invention” to innovation – where the market seeks to embrace a new product and rally to Foreword its access within the health ecosystem.

The 2022 “Global State of Digital Health Report” does far more than examine monies invested in the broad digital health category — it offers a guide for decision-makers on how multiple data points translate into market shifts and how to pinpoint trends that provide predictive insight and best practices in investment strategies, across regions, Digital Health categories, therapeutic areas and technologies.

Decisions will be made based on this collaborative effort. This report offers a new perspective – the data this report provides – finding the balance between hype and hope to offer balanced optimism is a guide.

The post Digital Health Innovation Continues to Press Forward at Warp Speed appeared first on Medika Life.

At a glance

Company Name: Zipline
Established: July, 2011
Company Type: Private, Series D, 233 Million USD
Field: Medical Supply and Logistics, Aviation, Robotics
Location: Half Moon Bay, California, USA
Official Website: FlyZipline.com
Vision: Our mission is to provide every human on Earth with instant access to vital medical supplies.

Core Product/Innovation/Service

Every now and then technology meets innovation and the result is a breathtakingly simple and effective product or service. Zipline may just be the best thing to hit healthcare since sliced bread. Fortune magazine calls the company a 21st-century solution to medicine’s “last mile” problem. It can save lives and deliver emergency medical supplies into remote and inaccessible areas. It works, it is currently being used with huge success in Rwanda and it’s starting to gain traction in remote areas near to you.

In a nutshell, Zipline delivers vaccines, blood, and other life-saving medicines into remote areas using drones. It’s the pharmacy version of Amazon taken to the skies in electrically powered small drones capable of operating almost anywhere and in climate or weather. 

The system works as follows. A doctor orders medicine using his phone, from a remote village based, in theory, anywhere. The village is inaccessible by road, but that isn’t a problem for Zipline. They despatch an autonomous drone, programmed with flight paths and local terrain information to the GPS co-ordinates of the doctor’s phone. 

On arrival at the GPS location, the drone descends to deployment height and drops its payload, a capsule or box attached to a small paper parachute. Minutes later the vaccine or medication is in the doctor’s hands and can be administered to the patient. It’s sheer genius and the best implementation of cutting-edge technology being deployed successfully into healthcare we’ve seen. 

Keller Rinaudo, the CEO and co-founder of Zipline explains the process in more detail.

Zipline has been rolled out in Rwanda and has proven itself, spectacularly so. Currently, a distribution center similar to the one pictured below can deploy more than a ton of medical supplies in one day. Zipline flies more than 40,000 km (over 25,000 mi) each day and they’ve logged over 6,000,000 km flown to real customers.

The little drones can operate in extremes: from the hot California desert to the rainy mountains of Rwanda. Zipline’s drones fly through real-world weather, day and night, carrying a payload of up to 1.75kg ((3.85 lb) each. That’s three units of life saving blood and multiple drones can be dispatched simultaneously to a set of coordinates.

Delivery of Covid vaccines to remote areas, quickly and without the need for extended refrigeration (they’re used as they arrive) is promising to open a whole new sector to these intrepid little aviators. Watch the video below to see just how amazing these drones are.

Meet the founders

Keller Rinaudo
CEO | Co-Founder
Email: None
Bio: Keller C. Rinaudo is an American robotics and autonomous airplane entrepreneur and the CEO and co-founder of Zipline. He was also the CEO and a co-founder of Romotive, a former company established in 2011 with Kickstarter funding that made inexpensive small robots that use mobile phones as their computing system, machine vision system, and wireless communication system. Romotive essentially shut down in 2014 and morphed into Zipline. Keller is a Harvard graduate (magna cum laude and Phi Beta Kappa with a 3.9 GPA) and grew up rock climbing. At one point, he was ranked in the top 10 in sport climbing
Connect: Linkedin | Twitter

Keenan Wyrobek
CTO | Co-Founder
Email: None
Bio: Roboticist Keenan Wyrobek’s first career was founding Willow Garage which created ROS (Robot Operating System) that now makes robotic components inter-operable. After creating a foundation to sustain this vital piece of open source software, Keenan went in search of new engineering challenges. He became intrigued with the idea of helping developing countries deliver crucial medical supplies to areas that are poorly served by roads. He co-founded Zipline, the VC-funded venture that is improving the lives of people in rural Rwanda. 
Connect: Linkedin | Twitter

Will Hetzler
Co-Founder
Email: None
Bio: Will Hetzler graduated from Harvard in 2009 as an economist. He worked as a consultant for Oliver Wyman for over three years, advising Fortune 500 aviation companies on engine maintenance, technical sourcing, supply chain and risk management.In 2014, he co-founded Zipline International in Silicon Valley to make drones to deliver critical medical supplies to remote parts of the world. He led the company’s work in Africa and established Zipline’s partnership with the Government of Rwanda. In 2018, he became the head of business development for the US and Canada.
Connect: Linkedin | Twitter

Connect with Zipline

The post Zipline: The New Gold Standard for Technology-Based Innovation in Healthcare appeared first on Medika Life.

Company Name: Marux
Established: 2019
Company Type: MARUX, INC. (Identification Number: 802516267)
Field: AI Healthcare Software Solutions
Location: 6001 Cass Ave, 2nd Floor WeWork, Detroit, Michigan, 48202
Official Website: marux.co
Vision: We envision becoming a default embedded solution for all emergency responders and medical personnel.

Core Product/Innovation/Service

Marux has developed a real-time AI-enabled remote monitoring app assisting medical care providers. Their solution offers a machine learning diagnostic application that streamlines critical care and seeks to address the following key issue.

Every 12 seconds a 911 emergency call is made. Thousands of lives could be saved by reducing emergency response times by just one minute. Faster medical diagnostics could save thousands more. Envision AI technology addresses these critical situations relaying vital diagnostic information to incoming emergency personnel and hospitals.

The solution seeks to improve response time in medical emergencies with instant availability of patient diagnostics and vitals, crucial in saving lives. MARUX™ can help save time in properly assessing and doing a pre-diagnostic on incoming patients.

It is achieved by real-time accident detection and instant AI-powered accident (health) data analytics in Marux cloud and real-time notification and streaming of data to first responders. Their CEO, Jonathan Lightning Rayos explains in the video below.

MARUX™ has three provisional patents in USA, Europe and China secured for their data process flow, integrated AI and real-time data capture software. They currently have commitment letters from international automotive suppliers for the use of their pilot program, as well as a scaled role out of their capabilities within a few key major metropolitan hospitals and EMS groups.

For more detailed investor related information explore these links or contect the company directly.

Webfunder, Company PDF

Meet the founders

Jonathan Lightning Rayos
CEO | Founder
Email: None
Bio: Has the superpowers of seeing into the deep future. Ability to design, code, transform and lead a team for digital transformation. Founded three technology companies in media, and hardware. Two successful exits & design patents. Self-taught in JavaScript, Python, Unity. Passionate and self-trained in golf, triathlons, snowboarding, and simulation racing.
Connect: Linkedin | Twitter

Michael ​D’Orazio
COO | Co-Founder
Email: None
Bio: With the superpowers to build and connect. Has vast expertise of centralizing teams of auto-mobility users. Founded a leading consulting firm for the OEM and Tier-1 sector in services, solutions, and systems applications. Passionate about Italian culture, classic cars, motorcycles, and family.
Connect: Linkedin | Twitter

Darko Stanimirovic
CTO | Co-Founder
Email: None
Bio: Over 10 years of experience in AR, XR, ML leadership. Master of Science degrees in Computer Vision + Robotics. Thesis in intelligence autonomous systems. Patent in ML, AR utilization. Has the brains and the brawn to lead designers and coders.
Connect: Linkedin | Twitter

Connect with Marux

The post Marux Offers an Innovative A.I. Solution for Emergency Responders and Medical Professionals appeared first on Medika Life.

Market valuations and forecasts paint a similarly optimistic picture, with projections showing that the AI healthcare market is expected to surge from $4.9 billion in 2020 to $ 45.2 billion in 2026, with a compound annual growth rate of 44.9 %¹. Unsurprisingly, the number of AI healthcare startups has also boomed, with 367 new start-ups receiving an estimated $4 billion in funding last year according to Accenture.

However, I am less certain about AI’s future in healthcare. The hype surrounding AI has reached such a furor that talks of “robo-surgeons” and “radiology replacements” have dominated public discourse in recent years, distracting us from evaluating the evidence behind these claims.

This article will demonstrate how the limitations of AI in healthcare have been brazenly glossed over with exaggerated claims, overpromises, and impractical scalability. It will highlight how we must significantly reduce our expectations of AI’s ability to transform tomorrow’s healthcare.

1. Data collection

Machine learning algorithms require a significant and comprehensive amount of labeled data input to classify data and predict outcomes with a high degree of accuracy. One way of obtaining this amount of data would be through obtaining electronic health records that integrate health and social care throughout the patient journey.

In the UK, an investigation into medical record-keeping in hospitals by the Institute of Global Health Innovation (IGHI) found that 23% of NHS Trusts were completely reliant on paper records³. That means 35 hospitals around the country have millions of patient data that remain locked and inaccessible.

Of the 77% of trusts that did use some form of electronic healthcare record-keeping, researchers discovered that systems were unable to record and share information effectively. This is highlighted by the fact that on 11 million occasions patients attended a hospital that could not access their full medical information from a previous visit².

Fragmentation of systems remains a significant barrier toward the implementation of AI. The critical information that is captured on the journey of patient care is, at best, inconsistent and, at its worst, unreliable. Over 4 million patients who attended two or more NHS trusts between 2017–2018, encountered hospitals with electronic record systems that were completely incompatible³.

Promises of Electronic Healthcare Records (EHR) that would create and sustain a wealth of data has clearly failed to materialize. Extracting quality data remains the bottleneck that severely inhibits AI integration with healthcare. Furthermore, the process involved to identify, organise and decodify the relevant data for real-life clinical practice makes AI seem a reality in a far, far distant future.

The failures of NPFIT, a government initiative that aimed to digitalise all medical records in the UK, failed so spectacularly that it was dismantled in 2011, being described as “the worst contracting fiasco in history”. It cost the government approximately £10 billion pounds.

Wearable technology and healthcare apps

It is true that wearable devices and applications may present a potential solution to quality data collection and interpretation. One study conducted by researchers at Stanford showed that the Apple Watch was able to detect irregular rhythms in wearers with a positive predictive value of 71%.

However, the main caveat to implementing this technology is security. Healthcare data obtained outside the doctor-patient relationship is akin to the Wild West, an open frontier poised for exploitation by bad actors.

Take the US federal trade commission, which tested 12 apps and two wearable devices. They found that data collected was transmitted to no less than 76 undisclosed third parties⁴. Perhaps more frightening is how period-tracking apps have shared the most intimate information on people’s lives, including when they had sex, to Facebook. This data sharing was performed without the informed and explicit consent of over 5 million app users⁵.

2. Algorithm bias

AI machines are thought to exist without prejudice, discrimination, and bias. Unfortunately, an increasing number of peer-reviewed studies have put that argument to bed. Consider the decision-making algorithm widely used by US hospitals to determine which patients receive additional medical care; the algorithm affected the care of approximately 2 million patients and researchers found it routinely discriminated against black people.

This study examined the data sets used in the algorithm and showed that “the care provided to black people cost an average of US $1,800 less per year than the care given to a white person with the same number of chronic health problems”.

The bias in the algorithm existed because it used healthcare costs as a proxy for healthcare needs, disregarding the fact that unequal access to care means there is less expenditure on black patients than white patients⁶.

Sourcing a diversity of data is difficult because minority groups have historically been underrepresented in clinical trials. This has been highlighted by the fact that in 2021, despite disproportionately higher infection, hospitalization and death rates in ethnic minority groups, the “direct effects of genetic and biological host factors remain unknown”.

Even if diverse data sets were used, bias can still arise through a lack of diversity in system developers, implicitly biased assumptions about the data being used, and the output data from AI itself, further perpetuating bias.

Using AI algorithms without extraordinary levels of due diligence risks deepening an already fractured society embellished with inequality and inequity.

3: Randomised controlled trials

Peer-reviewed evidence in randomized controlled trials (RCT) remains the gold standard for research. Yet, there have been few RCT’s for AI-based systems in healthcare. In addition to this, the overwhelming majority of RCTs that have been conducted are retrospective and have not been tested in real-life scenarios.

Prospective studies allow us to test and understand the viability of AI systems, as the performance of algorithms is likely to worsen significantly when the data being inputted is disorganized, chaotic, and different from the training algorithms used in practise.

To highlight this point, a systematic reviewconducted by researchers from Imperial College London, compared the performance of AI deep learning algorithms versus clinicians in diagnostic imaging. The results showed that only 10 records were found for AI RCT’s, and only 2 of them had been published.

In addition to this, 71% of papers analyzed had a high risk of bias, and reporting standards were suboptimal (<50% adherence for 12 of 29 TRIPOD items). Only one study produced the raw data and code.

To even begin to consider the possibilities of AI in healthcare, rigorous prospective RCTs must be conducted before any conclusion can be made about the effectiveness of such systems. Take this year alone, more than 3 vaccines are currently being distributed worldwide in the hopes of ending a pandemic. This would not have been a possibility without the stringent use of prospective RCTs.

And yet, the FDA has approved 16 algorithms for commercial use in the USA that affect millions of people’s health outcomes, and only one of these algorithms have undergone RCT testing⁷.

Using AI algorithms without reliable evidence is bordering on negligence. The use of robust, validated and transparent data has to be displayed in peer-reviewed studies before AI systems play a role in healthcare. Otherwise, the sensationalism reported in media could lead to the implementation of systems that are not fit for purpose, causing harm to millions of people.

4. Human factors

The adoption of new systems in healthcare is often met by a culture of resistance that is impervious to change. Problems with user-friendliness, accessibility, and slow speeds inhibit the uptake of technology that often sounds like a breakthrough in a sterile laboratory setting.

A study conducted by Google Health into the effectiveness of its detection rate of Diabetic retinopathy found that socio-economic factors affected the model’s performance in a clinical setting. This ranged from not enough lighting in a clinic room, to internet speeds being too slow, resulting in a large number of images being discarded.

Whilst human factors are often ignored when the implementation of new systems arise, they are perhaps, the most important to consider. There are numerous studies showing that failure to adopt promising technology boiled down to a failure to address the specific needs of the very people who will be using the technology. This is a costly mistake to make, given that millions if not billions of money may be spent in the future on the implementation of these systems.

In conclusion, we must stop this nebulous AI hype train that has derailed our expectations of what these systems can actually do. We are setting ourselves up for bitter disappointment when AI is unable to deliver a future that was over-promised. The lessons of IBM’s supercomputer, Watson, hailed as a transformation in cancer care failed to deliver on every single metric related to health outcomes. And yet here we find ourselves again, battling the same misinformation and hype that has characterized the last decade of our lives.

I leave you with the saying that has stood the test of time — all that glitters, is not gold.

References

1. Research M, Trends A. Artificial Intelligence in Healthcare Market with Covid-19 Impact Analysis by Offering, Technology, End-Use Application, End User and Region — Global Forecast to 2026 [Internet]. Reportlinker.com. 2021 [cited 27 January 2021]. Available from: https://www.reportlinker.com/p04897122/Artificial-Intelligence-in-Healthcare-Market-by-Offering-Technology-Application-End-User-Industry-and-Geography-Global-Forecast-to.html
2. Poor NHS medical record sharing is putting patient lives in danger [Internet]. Health Europa. 2019 [cited 27 January 2021]. Available from: https://www.healtheuropa.eu/nhs-medical-record-sharing-patient-lives-danger/95545/
3. Sandhu H. AI revolution in healthcare will occur when the tyranny of the screen ends — MedCity News [Internet]. MedCity News. 2020 [cited 27 January 2021]. Available from: https://medcitynews.com/2020/03/ai-revolution-in-healthcare-will-occur-when-the-tyranny-of-the-screen-ends/?rf=1
4. The Federal Trade Commission strikes back against Facebook [Internet]. Ft.com. 2021 [cited 28 January 2021]. Available from: https://www.ft.com/content/cd618d7a-bde8-4aec-86c7-f59ff784b4b4
5. Sex lives of app users ‘shared with Facebook’ [Internet]. BBC News. 2019 [cited 29 January 2021]. Available from: https://www.bbc.co.uk/news/technology-49647239
6. Obermeyer Z, Powers B, Vogeli C, Mullainathan S. Dissecting racial bias in an algorithm used to manage the health of populations. Science. 2019;366(6464):447–453.
7. Waters R. Human dimension presents new hurdles for AI in medicine [Internet]. Ft.com. 2020 [cited 29 January 2021]. Available from: https://www.ft.com/content/151dbf12-970d-49af-9d10-d417921f7066

The post No, Artificial Intelligence Will Not Transform Healthcare appeared first on Medika Life.