Fast and reliable application development and delivery is critical.This report exposes areas of breakage and disruption as many IT organizations continue to struggle to find cohesive approaches to application optimization.
Published Date: May 17, 2023
AI in healthcare is an umbrella term describing the use of cognitive technologies such as machine learning, natural language processing and robotics in medical settings. These technologies mimic specific behaviors associated with humans’ natural intelligence, such as the ability to learn from experience, extract meaning from language, and understand and move through a physical environment. They can be applied in healthcare to deliver better patient treatments, improve facility operations and assist healthcare professionals in their daily work.
Healthcare organizations accrue large volumes of data, including health records, medical claims information, images and clinical trial data. Like other industries, healthcare is increasingly managing these datasets in a cloud environment, which makes them available for real-time processing and analysis. Artificial intelligence capabilities such as predictive modeling, automation and deep learning help healthcare organizations deliver care more quickly and accurately — rather than relying solely on manual processes.
AI is a vast field with many potential applications in healthcare. However, healthcare organizations mainly rely on the following AI technologies:
- Machine learning: This branch of AI employs trained algorithms that enable computer systems to autonomously analyze and learn from data, perform specific tasks and improve their performance over time. In healthcare, it’s used for everything from optimizing administrative tasks to diagnosing illness to predicting the best patient treatments.
- Deep learning: Deep learning is a branch of machine learning that uses a model based on the brain’s structure called a neural network. This model emulates a system of human neurons, enabling it to draw conclusions from data much closer to the way a human would. When a standard machine learning algorithm makes an inaccurate prediction, a human must let it know and make necessary alterations so it can predict the desired outcome. Deep learning AI algorithms, however, can recognize the accuracy of their predictions on their own. This makes deep learning better suited to complex tasks — such as delivering personalized medical treatments, performing genomics analysis and understanding the effects of mental illness — than standard machine learning models.
- Natural language processing (NLP): Natural language processing is a branch of AI that enables computers to understand and interpret text and spoken language. NLP is used in healthcare to extract and analyze clinical information from patients’ medical records and lab reports, help doctors instruct robotic surgical instruments, and automate patient communications via automatic phone calls and chatbots.
- Robotics: Robots aided by AI technology such as machine learning and computer vision are popular throughout the medical field. They’re often used to support healthcare workers and enhance patient care. Trained robots, for example, can be used to transport medical supplies in a hospital or to safely disinfect and sanitize facilities, as they were during the Covid-19 pandemic. More sophisticated robots can also assist in patient treatments. Telepresence robots, which serve as virtual physicians in remote locations, and surgical robots that assist during surgical procedures, are examples of this.
- Robotic process automation (RPA): Robotic process automation, also called software robotics, doesn’t involve physical robots. Rather, it uses computer automation technologies to mimic administrative tasks such as form filling, transferring files and extracting data from records. It’s used in healthcare and hospitals to perform back-office operations like booking appointments, recording early authorizations, sorting and updating patient records, and processing bills and claims.
AI is an essential feature of healthcare organizations and hospitals, where massive volumes of data, expanding responsibilities and resource shortages can compromise the quality of patient care. In the following sections, we’ll look at how AI is used in healthcare and the benefits and challenges it brings to healthcare organizations. We’ll also look at how AI in healthcare has evolved and how the trends that are shaping AI will be used in future medical settings.
How is AI used in healthcare?
AI is used in healthcare in a variety of ways, including:
- Diagnostics: AI is used for more accurate and efficient diagnostics. Computer vision technology, for example, enables more accurate analysis of radiology reports and medical imaging, such as CT scans, MRIs and x-rays, as it can extract data that often isn’t visible to the human eye.
- Patient care: Healthcare organizations often automate patient communications through AI (sometimes via apps) to eliminate manual tasks such as appointment management, reminders and payments, allowing staff to devote time to more critical patient needs. AI also helps healthcare organizations connect disparate data to provide a unified picture of individual patients.
- Operations: AI technologies can quickly identify patterns in huge volumes of data (part of a healthy observability practice), helping healthcare organizations improve processes and workflows in virtually every facet of their operations.
- Training: AI in the form of “virtual patients” is increasingly being used to help clinicians and other medical staff hone skills ranging from performing procedures to explaining diagnoses.
- Research: AI technologies are used in disease and drug research to make these processes more effective and less costly.
What are some use cases for Healthcare AI?
Some popular use cases for AI in healthcare include:
- AI-assisted surgery: This technique combines the learning abilities of AI-based algorithms with the precision and control of surgical robots. AI-driven surgical robots collaborate with human surgeons during preoperative planning — analyzing a patient’s medical records to better pinpoint the surgical location, for example — and during surgery, by facilitating precise instrument placement and providing feedback, such as the size of a tissue. AI-assisted robots can also learn from past surgical data to inform new surgical techniques.
- Image analysis: AI-enabled imaging has proven to be faster and more accurate than manual analysis. A machine learning algorithm developed by an MIT-led research team, for example, was able to analyze 3D scans up to 1,000 times faster than previous techniques. This near real-time analysis has significant implications for surgeons, who could theoretically see whether a procedure was successful as soon as it was completed. AI-supported imaging can also help radiologists further support cardiologists, oncologists and other medical professionals to make more accurate diagnoses by analyzing electronic health records (EHRs) to identify patterns and relationships among thousands of images.
- Virtual nursing assistants: Virtual nurses are available around the clock to facilitate communication between patients and care providers. They can answer questions such as “What are the symptoms of Covid-19,” monitor patient conditions via medical IoT devices, provide clinical advice and follow up on treatments. By providing more regular communication between patients and healthcare workers, virtual nurses reduce unnecessary office visits and hospital admissions.
- Administrative automation: AI is widely used in healthcare to automate administrative tasks so healthcare workers can spend more time on patient care. It currently supports or performs back-office operations such as appointment booking, record management, billing and payer support. AI technology like speech recognition also supports physicians by helping order tests, filling prescriptions and processing patient notes.
- Aid clinical decisions: The use of AI helps medical professionals make clinical decisions by analyzing and correlating medical images, clinical research trials, medical claims and other healthcare data to identify patterns and insights humans might be unable to detect. This clinical decision support can improve health outcomes by helping clinicians predict illness and make faster, more accurate diagnoses.

While the list of use cases for AI in healthcare is extensive, some of the most popular include AI-assisted surgery, image analysis, virtual nursing assistants, automated administration and aid in clinical decisions.
How is AI in healthcare related to the Internet of Medical Things?
AI use in healthcare and the Internet of Medical Things (IoMT), are complementary technologies that enable healthcare professionals to monitor patient biometrics in real time and assist with diagnosis, treatment, patient recuperation and chronic care.
IoMT, also known as healthcare IoT, is the application of interconnected devices and advanced analytics in healthcare settings. IoMT devices connect patients, healthcare professionals and medical devices by transmitting information over a secure network.
IoMT expands the Internet of Things (IoT) into healthcare. Like IoT, it uses automation, sensors and machine-based intelligence to minimize the need for human intervention during routine healthcare procedures and operations. IoMT devices such as wearables, remote patient monitoring (RPM) devices, smartphones, point-of-care devices and kiosks, and in-hospital devices collect, manage, process and store medical data to be transmitted back to the healthcare organization’s network — which then will be carefully monitored.
AI technologies such as machine learning are applied to the data provided by IoMT devices to interpret readings, diagnose conditions, devise treatment plans, deliver therapies and monitor the patient over time. Although AI and IoMT technologies don’t eliminate the need for human involvement in patient care, together they dramatically reduce unnecessary doctor’s office visits and hospital admissions, ultimately lowering costs for both providers and patients.
What are the benefits of AI in healthcare?
Applications of AI in healthcare offer many benefits for both patients and healthcare providers.
- More accurate diagnostics: About 12 million people in the U.S. are misdiagnosed annually. Of those, around 10 to 20 percent are patients with serious conditions and 44 percent are oncology patients. AI helps overcome this issue by improving diagnostic accuracy and efficiency. AI technologies such as computer vision enable more accurate analysis of medical imaging and image formats such as x-rays, CT scans, MRIs and mammograms by extracting data not visible to the human eye. Machine intelligence is also able to analyze imaging data many times faster than humans, providing near real-time results.
- Better patient care: Hectic, overcrowded medical facilities are contributing to bad patient experiences. Poor communication is the worst part, according to 83 percent of patients in a recent study. AI streamlines communication by automating tedious tasks like appointment scheduling and reminders, bill payment issues and record management. AI can also rapidly process data, obtain and transmit reports and direct patients to the appropriate healthcare workers, allowing them to get what they need more quickly and efficiently.
- Improved clinical decision-making: AI solutions can quickly aggregate and process petabytes of clinical data to provide clinicians with a unified view of a patient population’s health status and provide near real-time access to data insights and predictive analytics that drive better patient outcomes.
- Safer surgeries: AI-enabled robots help human surgeons operate with greater precision and dexterity. As a result, patients experience reduced blood loss, infection risk, scarring and post-operative pain. Studies have found that AI-robot-assisted surgeries resulted in five times fewer complications compared to surgeons operating solo and a 21 percent reduction in the length of a patient's hospital stay.
- Improved preventative care and public health: Machine learning can be employed to process large volumes of medical, behavioral and environmental data to protect population health and predict disease outbreaks like the Covid-19 pandemic. AI solutions can then help healthcare professionals mitigate disease spread by rapidly diagnosing infected patients when they enter a healthcare facility and immediately enabling isolation and quarantine procedures. AI can also help speed up virus research and vaccine development. At the patient level, AI-powered wearables can help detect the signs of non-infectious health events and prompt the user to see a doctor before these conditions become serious.
- Increased healthcare accessibility: Studies reveal dramatic life-expectancy gaps between developed and underdeveloped countries. The latter typically lack the medical technologies and resources to deliver adequate healthcare to their populations. AI can enable digital solutions like telehealth to meet the increased demand for healthcare services due to events like the Covid-19 pandemic. It can also mitigate healthcare shortfalls in remote or underserved areas by facilitating remote diagnostics, treatment and monitoring.

Some of the many benefits of AI in healthcare include more accurate diagnostics, better patient care, safer surgeries and increased accessibility.
What are the challenges of AI in healthcare?
AI in healthcare presents a few challenges including:
- Limited visibility into AI models: Complex AI models are often needed to deliver better outcomes, but they typically work in a “black box,” meaning their inputs and operations aren’t visible to the user or are difficult to interpret. Black-box AI models can cause problems when healthcare workers need to understand how a model came up with specific results so they can take appropriate action. Ultimately, this lack of visibility can foster trust and reliability issues for both providers and patients.
- Diagnostic errors: AI solutions have been proven to increase diagnostic accuracy on the whole, but that doesn’t mean that every AI model will work as expected. Hundreds of AI tools were developed to detect Covid-19 during the first year of the pandemic, but none of them proved fit for clinical use. Diagnostic errors typically result from using poor quality data or making incorrect assumptions about the data, either of which will compromise the accuracy of the trained model.
- Lack of quality data: AI models are only as good as the data they are trained on. Unfortunately, the sensitive nature of medical data and the legal and ethical constraints around using it make it exceedingly difficult to collect. Even when you can collect enough training data, labeling all of it — which is required to make it recognizable and interpretable to the AI model — may be prohibitively time-consuming. Consider that a single imaging AI model may take 10,000 images to train, and the time and cost to build an accurate tool can become daunting.
- Data privacy: Patient data contains incredibly sensitive information, including biometrics, medical histories and payment information. Consequently, patient data is protected by strict regulations such as GDPR and HIPAA, with violations resulting in steep fines and other penalties. Unfortunately, patient data leaks and breaches are not uncommon; HIPAA Journal reports that there were more than 700 data breaches in 2021 alone.
These challenges may make some healthcare organizations reluctant to integrate AI into their systems. However, most can be overcome with education, effective tooling and appropriate security controls.
(Learn more about Splunk SOAR here.)
What are some healthcare AI risks?
The biggest risks of AI in healthcare include:
- Ineffective implementations: Despite AI’s many documented benefits, implementing and training AI solutions can be fraught with pitfalls for healthcare organizations. Because AI is a vast field, understanding it and how its various technologies and tools work can be challenging. AI models have to be trained with large volumes of quality data to produce accurate outcomes, and data can be compromised by anything from the method with which it was collected to the biases of the person inputting it. Even the best-designed AI solution can’t work on its own. Experienced healthcare professionals need a high level of training to work with their AI solutions to deliver effective patient care.
- Injuries and error: A poorly trained AI will be prone to error, which can result in harm to a patient. If an AI system fails to notice a tumor on a radiological scan, for example, the patient may not receive the necessary treatment, which could result in their death. Less dramatic AI errors could result in a patient not getting a hospital bed or having a medical record sent to the wrong facility. There’s also the potential for a problem in a single AI system to impact thousands of patients, far more than could realistically be injured by one provider’s error.
- Privacy issues: AI in healthcare requires the collection of large datasets from many patients, creating a significant privacy concern for healthcare organizations. Patient data is a high-value target for cybercriminals due to the amount of highly sensitive personal information it contains. And privacy is a concern even if a healthcare organization is never hacked — patients have filed lawsuits based on data-sharing between large health systems and AI developers, believing that collection of their healthcare data violates privacy. AI’s predictive capabilities can also create privacy concerns, as AI models make forecasts about patients regardless of the algorithm — an AI system might be able to identify that a person has Parkinson’s disease based on detection of hand tremors off of their wearable device, for example. This might be considered a violation of privacy, especially if that data were made available to a third party such as their insurance company.
What is the history of healthcare AI?
AI was introduced into healthcare in the early 1970s when a system called MYCIN was developed to identify blood infections such as meningitis and bacteremia. In the 1980s and 1990s, AI began to be used to collect and process data more quickly, enable robot-assisted surgery and implement electronic health records.
The inception of IBM's Watson AI system marked the beginning of a trend toward developing advanced systems that can rapidly and accurately answer questions. In 2011, IBM launched a healthcare-focused version of Watson that utilized natural language processing. Presently, other leading technology companies such as Apple, Microsoft and Amazon are also actively investing in AI technologies for the healthcare industry, joining IBM in this pursuit.
What is the future of AI in healthcare?
It’s impossible to know exactly what AI in healthcare will look like in the future, but we can get an idea based on how it’s being used today.
- Clinicians: AI will likely continue to improve diagnostic speed and accuracy, relying on more and better patient data to identify individuals’ health risks and propose preventative interventions.
- Healthcare administrators: AI will be key to streamlining processes, reducing bottlenecks and eliminating waste to help medical practices operate more efficiently.
- Patients: AI will facilitate more virtual healthcare options so patients can better manage health conditions on their own and get connected with healthcare professionals more quickly and efficiently when they need in-person care.
AI is profoundly influencing how healthcare is delivered. Although integrating AI into your healthcare system comes with many challenges, there are many more benefits for your organization and those you serve. AI can make your operations more efficient by automating mundane, repetitive tasks, streamlining workflows and getting users the answers they need when they need them. In all, AI helps you provide better, more personalized care and create a better experience for you and your patients.

Splunk Predictions 2023: Leadership Trends and Emerging Technologies
Splunk’s leaders weigh in on the most important technology trends facing the enterprise.