Like any technology, AI has just as much potential for harm as for good. Some experts predict that once the excitement and novelty of AI-assisted clinical procedures wear off, problems will begin to pop up. For example, few of the 130 AI devices the U.S. Food and Drug Administration (FDA) has approved over the past couple of years have been tested in clinical trials. As a result, AI could miss a tumor during a CT scan, recommend the wrong medication, give a hospital bed to a patient who needs it less than another and produce many other errors. And if there is a fundamental flaw in the programming, it could misdiagnose thousands of patients instead of just one.

Editor's note: This is the second part of a two-part series. The first installment can be found here. The first part installment in this series examined the benefits of Artificial Intelligence/Machine Learning (AI/ML) and noted the considerations that regulatory bodies are studying for use with AI/ML algorithms. The second and final installment explores the past and current regulations, and summarizes the latest framework proposed by the U.S. Food and Drug Administration's (FDA). While current guidance around AI/ML implementation in medical devices is lacking, the FDA is working to solve the problem.

In 2019, we saw increased interest and adoption of machine learning (ML) and artificial intelligence (AI) technology in healthcare. Organizations have been piloting solutions that range from helping diagnose patients, to ensuring the privacy of their data. While the industry is beginning to see some benefits from these tools, many end-users are starting to ask important questions like: how does the tool work, or where are my data stored? Similarly, in the last year, we have also seen organizations increasingly send and store their data at third-party vendors instead of on-premises. The combination of these two trends has raised concerns about data protection and the vendor's appropriate use of data.

In a first for the use of artificial intelligence in healthcare, the FDA has given a green light to a program designed to help primary care doctors diagnose autism in children at an early age, potentially when interventions may have the greatest effect on their neurodevelopment. Cognoa's Canvas Dx app takes in questionnaires filled out by caregivers and information from physicians as well as video of a child interacting with others or performing tasks, before providing assistance in making a diagnosis for autism spectrum disorder either within the doctor's office or remotely. Intended for children between the ages of 18 months and five years, the machine learning software aims to help make clinical decisions faster than previous methods--which may require referrals to a specialist plus a difficult process that can take years to identify varying symptoms. According to the FDA, though autism spectrum disorder affects about 1 in 54 children, and signs may first become apparent as early as 18 months, the average age at diagnosis may be closer to four-and-a-half years old. In addition, nonwhite children, females and people from rural areas or disadvantaged socioeconomic backgrounds are often diagnosed later or missed altogether, according to Cognoa.

Using a machine-learning algorithm, MIT researchers have identified a powerful new antibiotic compound. In laboratory tests, the drug killed many of the world's most problematic disease-causing bacteria, including some strains that are resistant to all known antibiotics. It also cleared infections in two different mouse models. The computer model, which can screen more than a hundred million chemical compounds in a matter of days, is designed to pick out potential antibiotics that kill bacteria using different mechanisms than those of existing drugs. "We wanted to develop a platform that would allow us to harness the power of artificial intelligence to usher in a new age of antibiotic drug discovery," says James Collins, the Termeer Professor of Medical Engineering and Science in MIT's Institute for Medical Engineering and Science (IMES) and Department of Biological Engineering.

AI, computer vision and machine learning systems proved that machines are better and faster than humans analyzing big data. Today, organizations have large datasets of patient data and insights about diseases through techniques like Genome Wide Association Studies (GWAS). Using AI, healthcare providers can analyze and interpret the available patient data more precisely for early diagnosis and better treatment. Today, it is possible to say whether a person has the chance to get cancer from a selfie using computer vision and machine learning to detect increased bilirubin levels in a person's sclera, the white part of the eye. As the interest in AI in the healthcare industry continues to grow, there are numerous current AI applications, and more use cases will emerge in the future.

Pediatric behavioral health company Cognoa has received FDA De Novo classification for its autism spectrum disorder (ASD) software diagnostic aid, Canvas Dx. The ASD diagnostic tool is designed to help primary care clinicians and pediatricians evaluate and diagnose suspected cases of autism among children. It uses machine learning algorithms to analyze videos of the child's behavior and responses to questions uploaded by parents and caregivers to devise a diagnosis. Canvas Dx is indicated as an aid in the diagnosis of ASD in patients between the ages 18 months and 5 years old who are at risk of developmental delay based on concerns of a parent, caregiver or healthcare provider. It is not intended to be used as a standalone diagnostic device.

Geoffrey Hinton is a legendary computer scientist. When Hinton, Yann LeCun, and Yoshua Bengio were given the 2018 Turing Award, considered the Nobel prize of computing, they were described as the "Godfathers of artificial intelligence" and the "Godfathers of Deep Learning." Naturally, people paid attention when Hinton declared in 2016, "We should stop training radiologists now, it's just completely obvious within five years deep learning is going to do better than radiologists." The US Food and Drug Administration (FDA) approved the first AI algorithm for medical imaging that year and there are now more than 80 approved algorithms in the US and a similar number in Europe. Yet, the number of radiologists working in the US has gone up, not down, increasing by about 7% between 2015 and 2019.

It's probably no surprise that money is pouring into life sciences and healthcare startups during the biggest medical crisis in a century. CB Insights reported that global healthcare funding hit a new record $31.6 billion in this first quarter of 2021. It's also no shock that the two biggest trends – artificial intelligence and telehealth – also reaped record amounts of private cash. AI healthcare startups raised nearly $2.5 billion, while telehealth companies did even better by netting $4.2 billion in equity funding. That's the third consecutive quarter to hit record highs in both sectors dating back to Q3'20.

The French artificial intelligence (AI) software developer Avicenna.AI has received clearance in the U.S. and the European Union for its Cina Chest AI software. Cina Chest, which includes tools for the detection and emergency triage of pulmonary embolism and aortic dissection from CT scans, has received the CE Mark in Europe and 510(k) clearance from the U.S. Food and Drug Administration (FDA). Cina Chest is part of Avicenna's Cina family of AI tools, which also includes the company's Cina Head software, which also has FDA clearance and the CE Mark. "Our pulmonary embolism and aortic dissection triage tools are the third and fourth algorithms we've released in less than 12 months, demonstrating our ambition to create AI applications that support detection and triage of emergencies throughout the entire body," said Cyril Di Grandi, co-founder and CEO of Avicenna.AI. The company was founded in 2018 and is based in La Ciotat, France.