AI in Healthcare
With the growing complexity and volume of data in healthcare, artificial intelligence (AI) will be used more often.
Updated
On this page
Updated November 2026. This was first published on 20 September 2021. The idea is the same. Any number below is what the original piece said then. It is not a new count.
Key takeaways
- With the growing complexity and volume of data in healthcare, artificial intelligence (AI) will be used more often.
- Several kinds of artificial intelligence (AI) are already being used by payers and providers of care and life sciences businesses.
- Diagnose and treatment recommendations, patient involvement, adherence, and administrative tasks are the most common applications.
With the growing complexity and volume of data in healthcare, artificial intelligence (AI) will be used more often. Several kinds of artificial intelligence (AI) are already being used by payers and providers of care and life sciences businesses. Diagnose and treatment recommendations, patient involvement, adherence, and administrative tasks are the most common applications. Although there are numerous cases where AI can do healthcare activities and or better than humans, setup issues will prohibit large-scale automation of healthcare professional positions for a long time.
Artificial intelligence (AI) and associated technologies are becoming more common in business and society, and they are starting to be used in healthcare. Many areas of patient care and administrative procedures within providers, payers, and pharmaceutical companies have the potential to be transformed by these technologies.
Several research studies have already shown that AI can perform as well as or better than humans in critical healthcare activities such as illness diagnosis. Algorithms are already beating doctors in detecting dangerous tumors and advising researchers to build cohorts for expensive clinical trials. But, we think that it will be several years before AI replaces humans in large medical process areas for several reasons.
Types of AI used in healthcare #
Artificial intelligence is not a single technology but rather a group of them. The majority of these technologies are immediately applicable to healthcare, although the particular procedures and activities assist greatly. Some critical AI technologies for healthcare are identified and discussed here.
1. Machine learning- Neural networks and Deep learning #
Machine learning is a statistical method for fitting models to data and learning from data through training models. Machine learning is one of the most prevalent types of AI. according to a 2018 Deloitte study of 1,100 US managers whose companies were already exploring AI, 63 percent of firms questioned were using machine learning in their operations. It is a comprehensive method at the heart of many AI techniques, and there are many variations.
The most frequent use of conventional machine learning in healthcare is precision medicine. predicts which treatment protocols are likely to be successful based on different patient characteristics and the treatment environment. The vast majority of machine learning and precision medicine applications require a training dataset for which the outcome variable (e.g., illness onset) is known. this is known as supervised learning.
The neural network is a more sophisticated type of machine learning that has been accessible since the 1960s and has been well established in healthcare research for many decades3 and has been used for classification applications such as predicting whether a patient would develop a specific illness. It considers inputs, outputs, and variable weights or ‘features’ that link inputs to outcomes. It has been compared to the way neurons interpret signals, although the comparison to brain function is poor.
There may be thousands of hidden characteristics in such models that are revealed by the quicker processing of today’s graphics processing units and cloud systems. Recognizing possibly malignant tumors in radiological pictures is frequent use of deep learning in healthcare.
2. Natural language processing #
Since the 1950s, AI researchers have sought to understand human language. NLP applications include voice recognition, text analysis, translation, and other language-related objectives. There are two approaches: statistical NLP and semantic NLP. Statistical NLP is based on machine learning (profound learning neural networks) and has recently improved recognition accuracy. It is necessary to have an enormous ‘corpus’ or body of language from which to learn.
The most common uses of NLP in healthcare include the production, comprehension, and categorization of clinical documentation and published research. NLP systems can analyze unstructured clinical notes on patients, produce reports (for example, on radiological exams), transcribe patient conversations, and perform conversational AI.
3. Rule-based Expert Systems #
Expert systems based on sets of ‘if-then’ rules were the leading AI technology in the 1980s and were extensively used commercially at the time. They were extensively used in healthcare for ‘clinical decision support’ reasons over several decades5 and are still frequently used today. Today, many electronic health records (EHR) providers provide a set of regulations with their systems.
Expert systems need to create a set of rules in a specific knowledge area by human experts and knowledge engineers. They function effectively up to a point and are simple to grasp. But, when the number of rules is high (often several thousand) and the authorities begin to clash, the system tends to break down. Also, updating the rules may be complex and time-consuming if the knowledge domain changes. They are gradually being supplanted in healthcare by more data-driven and machine-learning-based methods.
4. Robotic Process Automation #
This technology executes organized digital administration activities, such as those involving information systems as human users following a script or set of rules. Compared to other types of AI, they are less costly, easier to build, and more transparent in their behaviors. Robotic process automation (RPA) does not include robots but rather computer applications running on servers. To function as a semi-intelligent user of the systems depends on a mix of workflow, business rules, and ‘presentation layer’ interaction with information systems. In healthcare, they are employed for repetitive activities such as prior authorization, updating patient data, and invoicing. They may be used in conjunction with other technologies, such as image recognition, to extract data from faxed pictures and enter it into transactional systems.
Diagnostic and Treatment applications #
Disease diagnosis and Treatment have been a focus of AI since at least the 1970s when MYCIN was created at Stanford to diagnose blood-borne bacterial illnesses. This and other early rule-based systems showed promise for correctly identifying and treating disease but were not accepted for clinical practice. They were not much better than human diagnosticians, and inadequately linked with clinical workflows and medical record systems.
More lately, IBM’s Watson has gotten a lot of press for its emphasis on precision medicine, namely cancer detection, and Treatment. Watson uses a mix of machine learning and natural language processing (NLP). But, clients’ excitement for this use of the technology has waned as they recognized the difficulties of training Watson to treat specific kinds of cancer and integrate Watson into care procedures and systems. Watson is not a single product but a collection of ‘cognitive services’ delivered through application programming interfaces (APIs), comprising voice and language, vision, and machine learning-based data-analysis programs. Most observers believe that the Watson APIs are technically competent but that tackling cancer therapy was an excessively ambitious goal. Watson and other proprietary programs have also suffered from competition from free ‘open source’ programs offered by certain companies, such as Google’s TensorFlow.
Many healthcare companies are befuddled by AI setup difficulties. Although rule-based systems embedded in EHR systems are extensively used, notably in the NHS, they lack the accuracy of more algorithmic systems based on machine learning. These rule-based clinical decision support systems are challenging to maintain as medical knowledge evolves and are often incapable of handling the explosion of data and information based on genomic, proteomic, metabolic, and other ‘omic-based’ approaches to Treatment.
The future of AI in healthcare #
We think that AI will play a significant role in future healthcare services. It is the main capacity driving the development of precision medicine. is generally acknowledged as a much-needed improvement in Treatment. Although early attempts to provide diagnostic and Treatment suggestions have been difficult, we anticipate that AI will eventually master that area as well. Given the fast advancements in artificial intelligence for imaging processing, it seems probable that a computer will review adequately. Speech and text recognition are currently used for patient communication and clinical note collection, and their use will grow.
The most challenging issue for AI in various healthcare areas is securing their acceptance in everyday clinical practice, not whether the technologies will be competent enough to be helpful. AI systems must be authorized by regulators, connected with EHR systems, standardized to the point that comparable products perform similarly, taught to doctors, paid for by public or commercial payer companies, and updated in the field over time. These obstacles will be solved in the end. it will take considerably longer than the technologies themselves to develop. Consequently, we anticipate the limited application of AI in clinical practice over the next five years, followed by more widespread usage within the next decade.
It also seems apparent that AI systems will not replace human doctors on a broad scale but rather supplement their efforts to care for patients. Human doctors may eventually shift toward jobs and job designs that need distinctively human abilities like empathy, persuasion, and big-picture integration. Those healthcare professionals who refuse to collaborate with artificial intelligence may be the only ones who lose their employment over time.
Questions and answers #
What is the main point? #
With the growing complexity and volume of data in healthcare, artificial intelligence (AI) will be used more often.
What else did the original piece say? #
Several kinds of artificial intelligence (AI) are already being used by payers and providers of care and life sciences businesses.