Healthcare was less affected than other businesses by the decline in global AI funding this year because 2022 has arguably been the year of AI innovation across several industries, including healthcare, despite the economic slump.
According to the CBInsights research, financing for healthcare AI declined by 20% from Q2 to Q3, while funding for finance AI plummeted by 34% and funding for retail tech AI fell by over 50%.
As 2023 approaches, AI specialists are looking at 2022’s trends to forecast what to expect in 2023.
- An even bigger emphasis will be placed on personalised healthcare
According to a number of analysts, the industry will see an increase in the demand for tailored healthcare in the upcoming year. According to IDC, the amount of data in the world will expand fivefold between 2018 and 2025, from 33 zettabytes to 175 zettabytes. According to Morris Laster, partner for medical investments at Israeli startup capital firm OurCrowd, this rise in data will highlight individualised treatment.
According to Laster, “AI systems must be educated on enormous amounts of high-quality data in order to produce reliable predictions and suggestions.
He acknowledged that many healthcare datasets are incomplete, noisy, or biassed, which can result in erroneous or unreliable models. However, he claimed that businesses using AI in healthcare can overcome the issue by carefully curating and examining the data used to train their models.
He continued, “This may entail cleaning and preparing the data to remove any mistakes or discrepancies, as well as doing quality checks to guarantee that the data is representative and correct.
In order to improve patient results, Micha Breakstone, cofounder and CEO of Neuralight, predicts that the use of AI in healthcare will become more individualised. “Specific treatment can be developed based on the patient’s profile, genetics, surroundings, and lifestyle,” he says.
According to Andy Thurai, VP and primary analyst at Constellation Research, AI can be useful in the field of precision medicine or personalised medicine.
Because of the data needed to develop this personalised profiling, he claimed that up until recently it was nearly impossible to configure the treatment, medication combination, and drug mixing that can work for a certain patient. But with today’s advances in genomic profiling, he explained, AI can instead of relying on a pharmacist or doctor, determine what treatments will work and what won’t based on a person’s genetic profile, past treatment history, or current medication history and develop a customised treatment for every patient.
The availability of wearable medical devices, telehealth consultations, and predictive diagnosis using gathered data in real time can all lead to identifying the right person for treatment while taking unnecessary treatments out of the picture, Thurai said. This can free up the time of the doctor and other healthcare professionals to [spend] where it is needed. Additionally, wearable technology and AI that escalates the issue only when essential to set up an instant visit have made remote health monitoring simpler.
- Healthcare AI legislation and regulations will advance
In the healthcare sector, a lot of data is produced. The industry “produced 2,000 exabytes of data in 2020 and will continue to increase at a 48% rate year over year,” according to one IDC assessment.
This presents particular data issues to the healthcare sector, particularly in regards to data protection. McKinsey examines whether legislation will benefit the AI sector in the same manner that the General Data Protection Regulation (GDPR) has benefited privacy protection, as well as the extent to which regulations will alter approaches to ethics, health data, and patient confidentiality.
However, Ayanna Charles, a solutions consultant for the manufacturer of predictive software Verikai, thinks that laws and regulations will improve over the course of the following year. Legislative and regulatory organisations are becoming more aggressive in their regulation of data sharing and the use of AI in healthcare, she added, both at the state and federal levels. Large insurance and healthcare organisations are by nature conservative, so we anticipate greater direction on what constitutes appropriate usage of data and AI.
Charles continued by saying that throughout time, the amount of data generated will increase, increasing the usefulness of AI. Additionally, she predicted that it will lead authorities, business, and advocacy organisations to “coordinate around a uniform framework and set of procedures that balance the need to protect patients’ data with the practical medical benefits of using that data within AI models.”
- The fight against AI prejudice will receive increased attention.
While bias in models is a major problem that must be addressed by any industry employing AI, a Harvard study finds that healthcare professionals must pay particular attention to this issue.
According to the report, “Biases in healthcare AI can further increase social inequality and can cause death.”
However, Charles pointed out that biases for AI in healthcare will start to decline starting in 2023.
Before developing and deploying any models, she stated, “anyone employing AI in healthcare must address the bias that exists inside the health system as it exists today.” “At worst, people might construct and use a model without thinking about it, which would spread and reinforce preexisting bias. But at their finest, AI models can be employed to minimise and eliminate inconsistencies in the healthcare system.
According to Lester, organisations employing AI in healthcare can put in place policies and supervision procedures to address ethical concerns and make sure their AI systems are being utilised ethically and in accordance with the law.
In addition, responsible AI “helps achieve fairness, even though biases are baked into the data; gain trust, even though transparency and explainability methods are evolving; and ensure regulatory compliance, while grappling with AI’s probabilistic nature,” said Svetlana Sicular, research VP at Gartner.
- More uses in the healthcare sector
Accenture predicts that by 2026, AI applications will reduce healthcare spending in the United States by $150 billion annually. These savings are anticipated as a result of more applications. Tzvi Bessler, a medical analyst for OurCrowd, predicted that over the next five years, starting in 2019, AI will continue to play an increasingly significant role in the healthcare sector.
According to Bessler, “this will probably require the development of more advanced and clever AI systems that can make more precise and trustworthy forecasts and suggestions.”
According to Laster, AI will be employed more for jobs like medication discovery and development as well as for enhancing the effectiveness and precision of medical research in 2023.
Thurai stated that he also thinks AI will significantly contribute in the discovery of new medications.
He declared, “This has already begun, and there will be more next year.”
In a circumstance like the most recent COVID-19 pandemic, he continued, “drug and vaccination development and research in clinical trials cannot take the regular lifetime given the immensity of impacted persons and the pace at which the virus disseminated.”
A vaccine “can take years to bring to the market, but the COVID vaccine was mass-produced in billions and saved many lives using AI-assisted features such as discovery, experimentation, side effect, and efficiency tracking and so on — which would have been impossible to track using normal methods,” the author said.
- A tighter collaboration between AI and people
Despite claims to the contrary, Charles doesn’t believe AI will ever completely replace humans, at least not in the healthcare sector.
She instead envisions a closer connection between people and machines.
According to her, “AI technology should not be considered as the sole solution to the business problem facing an enterprise, as with all new technologies.” “Technologists must create and maintain a robust governance framework before implementing any AI solution; this structure must combine the technology with human oversight, uniform rules, and reliable processes.”
In fact, according to Gartner, “all individuals employed for AI research and training activities will have to demonstrate expertise in responsible AI, heralding a closer link between humans and computers” by 2023.
Thurai continued by saying that there must be simple and useful methods for overturning AI judgements.
Many managers and executives who are already using AI acknowledge that they have had to interfere with their systems because they delivered inaccurate or unfair findings, according to Thurai. “In the coming year, we will start to see corporations giving a better human-AI interaction a higher priority, especially in an industry like healthcare,” says one expert.
- More automation, more automation
2023 will see increased automation in the healthcare industry, according to Laster.
He pointed out that we should anticipate seeing AI utilised more frequently for jobs like keeping patient records, making appointments, and coordinating treatment.
According to him, this might increase the effectiveness and efficiency of healthcare delivery while giving patients access to more convenient and individualised care.
According to a recent McKinsey study, “AI-enabled personal assistants can automate 50 to 75% of manual chores, enhancing productivity, cutting costs, and enabling doctors to focus on challenging patients and actual care delivery and coordination. In consequence, this might enhance both the clinical and insurance plan members’ healthcare experiences.
A bright future for AI in healthcare
Thurai acknowledges that it will take some time for AI systems to be able to reflect the empathy that influences many human decisions, but that doesn’t mean they shouldn’t be continually refined to more closely resemble human ideals.
Business leaders need to be aware that the decision-making process is more complex than just using cold, data-driven insights, he said, adding that AI simply reflects the programming and data that go into it.
