The i-value

HPHR Fellow Javaid Sofi

By Javaid Iqbal

Role of Machine Learning in Healthcare

Artificial Intelligence and Machine Learning have emerged as one of the most important technologies of Industrial Revolution 4.0. The use of AI and ML will transform and disrupt healthcare as large quantities of data will be transformed into insights for improving the diagnosis of treatment and public health responses, and rapid drug discovery. According to McKinsey, AI and ML will deliver an economic value of $13 trillion by 2030. Gartner expects the global AI-based economic activity to increase from about $1.2 trillion in 2018 to about $3.9 trillion by 2022.

 

Health care is perhaps the most complicated industry on earth. It involves science, business, politics, and volatile human behavior. Data and human judgment have always been involved in healthcare; with improvements in AI and ML, they are getting closer. These developments are changing the way providers make clinical decisions with precision in diagnosing, treating health management. From genomic and phenotype data to delivering systems, AI generates enormous amounts of data that can be harnessed to improve the quality of care.

 

However, AI alone cannot transform healthcare. An algorithm widely used in US healthcare organizations has been found to discriminate against people of color systematically. The research found only 17.7% of patients assigned received extra care when the actual number should have been 46.5%. 

 

Around 250,000 deaths per year occur due to medical error in the US, the third leading cause of death after heart diseases and cancer. Machine learning and AI can help us avoid these errors and deaths. It also can decrease the managerial burden of physicians by improving clinical decision software. With natural language processing, notes can be converted into Electronic Health Records, which means a physician has to enter the data only once. AI-enabled software can also provide access to data from multiple sources—including smartphones and connected medical devices, which can transform health care and create personalized care delivery. 

 

According to the American Medical Association, every year, the healthcare industry wastes up to $935 billion, around 25% of the medical spending in the US. According to Insider Intelligence, 30% of healthcare costs are associated with administrative tasks. Tasks such as insurance, following up on bills, and maintaining records can be easily automated. Hospitals and insurance companies use algorithms to help manage care for 200 million people in the US. There has been an 88% increase in the number of organizations that have implemented an AI strategy compared to the previous year.

 

With these improved technologies, patients may not need many expensive tests, bringing down their healthcare expenses and making them more accurate. For instance, according to the American Cancer Society, 12 million mammograms are performed yearly in the United States, but half false yield results. One in two healthy women is misdiagnosed with cancer. AI enables the review and translation of mammograms 30 times faster with 99% accuracy, reducing unnecessary biopsies. 

 

Research has shown AI and ML can help in the early detection of Heart diseases and skin cancer. At Indiana University-Purdue University Indianapolis, researchers developed a machine-learning algorithm to predict (with 90% accuracy) the relapse rate for myelogenous leukemia (AML). Spending on healthcare is expected to rise by 4% yearly between 2020 and 2024, Machine Learning and AI can help reduce the amount of money people spend on healthcare. Robotic surgery fueled by AL and ML algorithms has allowed surgeons to control robotic limbs and perform surgeries with added accuracy and minor tremors. According to Accenture, robotics has reduced the length of stay in surgery by almost 21%. 

 

The use of AI chatbots can aid in interactive and private communication, which will help populations suffering from stigmatized diseases like HIV/AIDS or any mental issues. AI solutions are also used to identify new potential drugs or therapies from vast information databases about the current medicines, which help bring new drugs quickly into the market. 

 

Most models look backward, yet they are asked to make decisions that’ll impact us in the future; AI and machine learning help make data-driven decisions in real-time and for the future. Machine-learning-optimized laboratory testing at MIT has fast-tracked detection of different antibiotics earlier considered unachievable due to the significant time and financial investment.

 

We have to make sure everyone understands the risks and capabilities associated with AL and ML. We need social change that includes making inclusive policy changes around AI and automation. We need to partner with different stakeholders, including government, academia and private industry, and non-profit organizations. Systems that are not appropriately designed will misdiagnose, and algorithms trained on data sets with biases will incorporate those biases. 

Like what you read?

More from Javaid Iqbal here.