AI Proves Effective at Diagnosing Diseases: What’s Next?

AI Proves Effective at Diagnosing Diseases: What’s Next?

By: TEAM International | October 21, 2020 | 19 min

These days, 94% of healthcare executives state that their organizations have accelerated the pace of innovation over the past three years, thanks to AI, RPA, and ML technologies. Meanwhile, the market for AI in healthcare is projected to reach $6.6 billion by 2021. On top of that, the COVID-19 pandemic has pushed us to transform almost every industry into all things virtual. Especially – the medical field. With this in mind, it seems like now is the best time to invest in healthcare digitalization, but is it truly worth the cost?

While pharmaceutical companies worldwide are racing to create a vaccine for COVID-19, software development providers also continue contributing to the life sciences sector and artificial intelligence in medicine in particular. They create AI and RPA solutions that help medical staff work faster and with higher accuracy.

However, people have been debating about entrusting the healthcare system’s future to AI and IoT for years. Yes, we saw ‘Black Mirror,’ we may guess what you think. Let’s sort this out and see what awaits us in the nearest future when we say, “AI is good for global adoption.”

Is AI really better at diagnostics than humans?

Cancer

In 2020, DeepMind and Google researchers trained an AI tool for early detection of breast cancer using machine learning algorithms and microscopic specimen images. The accuracy of the results was in the range from on a par to outperforming the medical experts. So, Google AI beats doctors at scanning the mammograms.

The reason for this outcome is the human factor – pathologists and radiologists can get tired, distracted, or lack the needed expertise. It’s horrifying what disastrous consequences one misdiagnosis can have from prescribing unnecessary treatment to wrongly established all-clear.

Even an excellent doctor can make a life-altering mistake being affected by some external factors. In contrast, artificial intelligence software doesn’t get tired and processes loads of gigapixel images effortlessly. It learns and self-improves from the materials humans fed to it.

Diabetes

Google deep learning tech also taught the neural networks to detect diabetes at early stages based on deviations in patients’ retinas photos. This algorithm similarly proved to be a bit better than human ophthalmologists.

Tuberculosis

Two scientists from the Department of Radiology, Thomas Jefferson University Hospital, Sidney Kimmel Jefferson Medical College, conducted a study that showed that deep learning in healthcare could identify pulmonary tuberculosis. The technology processed patients’ routine clinical data, and the results were set against human predictions that followed the American College of Cardiology/American Heart Association (ACC/AHA) guidelines. As it turned out, AI correctly predicted 7.6 percent more heart attacks and showed 1.6 percent fewer false alarms than cardiologists via the ACC/AHA method.

Brain strokes

Brainomix (UK) develops an AI product able to analyze brain scans of people with a suspected stroke. Timely and correct detection of blockage in the blood supply to the brain may save patients’ lives.

The Brainomix software analyses CT brain scans in one minute, whereas medical practitioners require much more time and often miss the critical signs. Based on successful trials, 20 NHS hospitals have already implemented this technology in their work routine.

Eye diseases

The coalition of DeepMind, Moorfields eye hospital NHS foundation trust, and University College London created a disruptive AI-based healthcare system. Reading eye scans it can correctly detect more than 50 different eye diseases and guide patients to further treatment with 94 percent accuracy. Such results both match and beat the world-leading ophthalmologists’ diagnoses.

All-in-one

Several scientists published a study in The Lancet Digital Health, where they compared performances of deep learning tools versus medical professionals in detecting diseases from medical images. Their first of a kind systematic review and meta-analysis of 14 compared studies within the same sample showed that AI and healthcare workers’ performance indexes were practically equal.

The deep learning systems’ accuracy rates were 87 percent for disease detection and 93 percent for all-clear diagnosis. Humans made correct diagnoses for these cases with 86 and 91 percent, respectively. It means AI can be as effective as clinicians at diagnostics. Such eHealth systems may prove extremely useful for hospitals lacking medical specialists able to read medical imaging since they can speed up the diagnosing process.

Health AI market size 2014-2014

The use cases of AI in healthcare: 2020 and beyond

We’ve already briefly discussed the current rise of virtual care, telemedicine, the Internet of Medical Things, and automation as the catalysts for the future of healthcare. Today eHealth uses deep learning algorithms in pathology, radiology, oncology, ophthalmology, cardiology, and other areas.

AI subdomains used in eHealth software

Applications of artificial intelligence in the medical field

AI in healthcare can accurately identify the linkage between genetic codes, discover rare genetic disorders and a range of syndromes, guide surgical robots, and more. How is this tech disrupting the healthcare areas here and now?

Administrative

AI eliminates the waste of healthcare workers’ time and resources, increasing their efficiency in administrative load and regulatory compliance. Chatbots and RPA tools optimize and streamline healthcare operations, patient care, billing and supply chain administration data privacy, controlled substances management, and more.

They can make the document flow error-free and fast, improving the operational outcome for health facilities. Physicians will be able to spend more quality time with patients instead of dealing with paperwork.

Companies to watch: Olive, Notable Health, Alpha Health, Qventus, Protenus, HealthVerity, Ribbon Health, Vault Medical Records, Cerner, Augmedix, Epic Systems Corporation, Suki, KenSci.

Patients Care & Treatment Delivery

AI solutions can make primary care financially affordable and accessible to all people. These systems can improve patient experience and engagement. The natural language processing (NLP) subdomain provided us with human-like interfaces that automate patient screening, care navigation, patient-doctor communication, prescription audit, and more.

Companies to watch: Babylon Health, Memora, Tempus, CloudMedX, Gyant, Enlitic, Zakipoint Health, Curai, Wellframe, H2O.ai, ICarbonX, Oncora Medicals, Jvion.

Early detection

AI and machine learning in healthcare ensure earlier and often more accurate disease detection by scanning complex medical images and leveraging healthcare data. Patients will get less expensive diagnosing and testing services.

Companies to watch: IBM, Proscia, Caption Health, PathAI, Paige, Huiying Medical, Ezra, Zebra Medical Vision, SkinVision

Diagnosis

Five years ago, misdiagnoses and human errors accounted for 10 percent of all deaths in the USA. When it comes to AI-based healthcare solutions, their capability to accelerate the diagnostic process, reduce errors, and thus save lives looks promising. They have the potential to predict and detect diseases quicker and with higher accuracy than most medical experts. Learning algorithms can suggest the most effective treatment options, improving patients’ outcomes. AI-enhanced microscopes can scan for harmful bacteria in blood samples faster than if you do that manually.

Companies to watch: Google, DeepGestalt, PathAI, Zebra Medical Vision, Freenome, Buoy Health.

Decision-making

Artificial intelligence in medicine helps humans boost the diagnostic process with timely decisions based on the collected and aligned big health data. Collecting and synthesizing patients’ data from EMRs, bedside monitors, and other medical devices, AI provides predictive analytics that supports and empowers clinical decision-making. For instance, computer vision monitors blood loss during childbirth with a higher accuracy rate than most clinicians.

Companies to watch: Gauss Surgical, Medical Informatics, DeepMind Health.

Telemedicine

During the pandemic, we’ve witnessed the rise of AI-based virtual care tools for preventive care and ongoing support with no need for in-person physician visits. Now those patients who are not in life-and-death situations can receive expert clinical services remotely via a video call or chat with a healthcare professional. Chatbots are also becoming conversational assistants built for specific use cases that close the gap between physicians and patients.

What’s more, artificial intelligence recognizes the signs of depression, risk of domestic violence, or suicidality in your speech and texting patterns. Smart apps and therapy programs play a significant part in providing broader mental healthcare access and reducing substance abuse.

Companies to watch: Eko, Aluna, Biofourmis, Current Health, Myia.

Pharmaceuticals (R&D of new drugs)

Medical institutions worldwide team up with tech companies to design AI products that will help develop and test new drugs for rare diseases treatment faster and more cost-efficiently. The use of AI in drug development means creating much cheaper precision medicine tailored to patients’ needs based on genetic and molecular data.

Companies to watch: Bayer, Deep Genomics, Syapse, BERG Health, BenevolentAI, Atomwise, GNS Healthcare, XtalPi, NuMedii, BioXcel Therapeutics, 4Quant.

Surgeries & training

Surgeons have been actively using augmented and virtual reality technologies to train and perform complex surgeries for years. Robots have evolved into full-time assistants in laboratories and operating theaters. Moreover, they are even used to provide end-of-life care (EoLC) for the elderly. With deep learning in medicine, you can tune hearing aids and sophisticated ultrasound machines. The exoskeletons and bionic prosthetics get better and better with each passing year.

Companies to watch: Intuitive Inc., Auris Health, Vicarious Surgical, Accuray’s Cyberknife System, Mazor Robotics, MicroSure, Carnegie Mellon University.

A twilight zone: how do the AI tools blend with real people?

The hype around artificial intelligence in the medical field has raised concerns that someday the technology might replace physicians. However, we think it’s impossible, and more likely, AI is just going to augment the roles of healthcare workers, complimenting their practices. Wider implementation of this technology will create new career opportunities as healthcare staff is expected to become more tech-savvy. People can achieve so much more working hand in hand with intelligent automation tools.

Artificial intelligence as a silver lining for healthcare workers

The COVID-19 pandemic has proved that the main course of healthcare executives is not only to fight against the unknown disease but to create a strong protection basis for their medical staff.

AI will never completely dehumanize the medical industry. Instead, if applied appropriately, it can protect medical workers with digital monitoring and portable eHealth devices. AI systems are the silver lining for healthcare workers that have accelerated remote care and diagnostic adoption — telemedicine. In the future, deep and machine learning algorithms can become fierce warriors standing on the frontline to reduce doctors’ and scientists’ exposure to patients with COVID-19 and other life-threatening infections, viruses, hazardous substances. They can also add a layer of decision support in the most critical situations.

Being in business for almost 30 years, a reputed development partner like TEAM International knows how to navigate the challenges and create relevant AI and RPA solutions for the eHealth industry to deal with the most complex problems. We can conduct an audit of your processes and systems to identify where AI and cognitive automation will be the most effective.

The revolution of AI and machine learning in healthcare: what’s next?

top 10 AI applications

Afterword

Learning algorithms are becoming more precise as one trains them by feeding with new data, diagnostics and treatment insights, and more. The use cases of AI in healthcare reviewed above prove that the algorithms can perform either better than or as good as physicians when it comes to diagnoses and predictions. We can only guess how much better than humans these algorithms will be in five years. It’s just the beginning of the AI-based revolution. We’re on the verge of witnessing dramatic changes in the eHealth industry.

Latest Industry Insights