healthcare worker at computer with robot

How Automation is Transforming Healthcare and Saving Lives

Nahla Davies

Automation anxiety describes the fear associated with losing your job to machines. Major discoveries and advancements in artificial intelligence have become more frequent recently. Automations powered by artificial intelligence are able to rapidly improve productivity and accuracy on routine tasks. Yet, the concept of automation can arouse a sense of uncertainty and fear. Concerns are generally with accuracy, privacy and job loss when automation is used.

The pandemic illustrated the massive importance of modernizing healthcare through automation. Sophisticated scheduling software streamlining vaccination programs and AI assisting in contact tracing are only a few examples of the positive impact of automation on healthcare. The following guide will examine other ways that automation has revolutionized medicine, saved lives and why it should be embraced.

Levels of Automation in Medicine: An Overview of Current and Future Automation in Healthcare

Before we can understand automation’s impact on healthcare, we must understand all facets of its current and future implementations. According to Dr. Bertalan Mesko (MD, Ph.D.), there are at least five levels of automation in medicine: human only, shadowing, AI assistance, partial automation and full automation.

Level 1: Human Only

There is virtually no artificial intelligence involved in the first level of automation. It is essentially traditional medicine where most procedures are performed manually by a licensed practitioner. The practitioner may consult AI or algorithm-driven tools such as medical search engines. However, diagnostics and treatments are mainly carried out by the physician. This is the automation paradigm of the past, where medical professionals are at the forefront of healthcare with very little assistance from automated systems.

Level 2: Shadowing

Medical students are often assigned to an experienced physician in medical school. This practice of “physician shadowing” gives students the opportunity to observe the physician-patient interaction while gaining practical experience in the medical field.

Artificial intelligence is also capable of performing a variety of shadowing where it observes a medical professional conducting a diagnosis or a procedure. The AI will shadow the physician without influencing the procedures.

The AI’s primary objective is to accumulate and log as much data as possible. This data can then be used in the future to determine if and why certain treatments are more effective than others. Furthermore, it allows physicians to analyze the data and identify any mistakes in how they conduct procedures. This will help minimize malpractice.

Level 3: AI Assistance

At this level, the AI provides guidance and supports the medical practitioner in diagnostics and clinical decision-making. The AI may make suggestions by using past data and evidence to assist the physician during the diagnostic process.

IBM’s Watson Health is a good exampleof this. It aids oncologists by providing faster and more accurate cancer detection and treatment. Watson Health mainly does this by amassing a large collection of data on the latest research. Additionally, it scans through the patient’s history to identify the most viable treatment solutions.

There are challenges in ensuring that the recommendations are accurate and easy for practitioners to use and interpret. Yet, the data shows large improvements of accuracy when AI is used.

Level 4: Partial Automation

During partial automation, the AI will produce its own diagnosis. However, it may flag cases and consult physicians if it encounters any uncertainties. This provides can free physicians for other tasks while still ensuring that complex cases are addressed by them.

In 2019, a team of medical investigators from the Massachusetts General Hospital’s Department of Radiology developed and tested an AI system that could rapidly detect and categorize brain hemorrhages.

The system produced many promising results, such as 93.0% diagnostic accuracy for main types of intracranial brain hemorrhaging (ICH). However, the results were less desirable for specific subtypes of ICH. Nevertheless, a system like this (after some refinement) can be used by radiologists to detect ICH quickly and potentially save lives.

Level 5: Full Automation

This level of automation is what most AI and machine learning scientists in healthcare envision for the future. This is where healthcare processes are performed by an AI without any human input. For instance, a fully automated diagnostic system could analyze a CT scan and initiate subsequent testing. It would perform the entire procedure without consulting a human physician.

While attaining this level of automation is ideal, many experts believe that it’s nigh impossible to achieve anytime in the near future.

The Holistic Potential of AI-Assisted Medicine

The proposed five levels of automation provide a succinct overview of automation in the medical industry. From this perspective, focus is mostly centered on making the lives of professionals easier. This will no doubt improve patient care too. However, it is still imperative that we address all the tangible ways automation can assist in patient care and ensure consistent treatment.

AI and Automation in Mental Health Care

The mental health implications of the Covid-19 pandemic and its subsequent fallout have been well documented. According to a study conducted and published by the Kaiser Family Foundation, 4 out of 10 American adults reported symptoms of anxiety and/or depressive disorder.

Shutdowns, curfews and other restrictions made maintaining treatment for mental health a challenge. There has always been a seemingly underreported mental healthcare crisis looming. How can AI and automation help?

One of the biggest challenges that has always been present in mental health diagnosis is establishing objective metrics to identify disorders and illnesses. Historically, practitioners have relied on subjective approaches to discover what ails the patient and determine which treatments are the most suitable. This is more of a troubleshooting approach to treatment and could be improved with the assistance of AI.

Mental health professionals can now use breathing, speech patterns, typing patterns on their devices, physical activity, and social interactions to diagnose their patients and recommend treatments. All this data can be collected from IoT devices and AI-driven diagnostic tools. See a few ways that IoT is changing healthcare.

However, AI technologies' contributions to mental healthcare aren’t limited to diagnoses. They can also assist in treatment. Edith Cowan University found that 30% of people are more comfortable sharing negative experiences with virtual reality avatars than they are with actual people.

Therapy has long been inaccessible for many patients. VR, in conjunction with AI, can help people overcome this barrier while providing more accurate diagnoses.

AI and Automation in The Treatment of Diabetes

There are many challenges related to diagnosing and treating diabetes globally. According to the CDC, 1-in-10 Americans (over 37 million) have diabetes. However, 1-in-5 may not know that they have it. Furthermore, they may never get diagnosed or treated.

Diabetes is a massive public health problem and one of the leading causes of blindness (through diabetic retinopathy) in the US for people aged under 75. IDX-DR is a level 4 autonomous system that can analyze retinal images and make diagnostic decisions based on them. Paired with the latest optometry software, this can provide early preventative treatment against diabetes-related blindness.

However, this is only one way that AI-powered technology is revolutionizing diabetes treatment. Continuous glucose monitoring (CGM) systems are another way. These are medical devices that sit semi-permanently under a patient’s skin. The patient can then use an external wireless device such as their smartphone or smartwatch to check their glucose levels before administering insulin.

Telemetry and health data gathered from these devices have the potential to be used by insurance companies to form life insurance policy quotes. This could allow them to streamline costs. Additionally, the CGM can send reminders or alerts to the user regarding insulin (as well as other medications such as metformin) intake.

Diabetics are more at risk of passing out from hyperglycemia and hypoglycemia. The CGM system can automatically signal emergency health care services if it detects that you have lost consciousness abnormally.

The most popular and pragmatic applications of machine learning and AI have always been in data collection and sorting. As more people embrace IoT and health-related smart devices, such as fitness trackers, the greater the potential for early detection. This allows healthcare professionals to collect more accurate information related to patient behavior. This can improve the early detection of diabetes and other diseases.

However, human operators cannot effectively sort through this information manually. That’s why we employ machine learning to capture and find relevant data to assist medical professionals in making informed decisions.

AI and Automation in Pharmaceuticals

Many pharma companies had already begun employing Artificial Intelligence and machine learning to improve their drug and vaccine production protocols even before the COVID-19 pandemic. However, AI’s potential was magnified during the pandemic.

Vaccine discovery that typically takes over a decade was fast forwarded thanks to AI-powered technology. A fine example of this is AstraZeneca.

AstraZeneca is one of the largest pharmaceutical companies in the world. Incidentally, it has begun implementing AI and data science-driven tools across its research and development fronts.

The company is using machine learning-powered image analysis and knowledge graphs to collect important insights about diseases. AI has been shown to identify markers 30% faster than human pathologists. 

Additionally, AstraZeneca has recently released a few AI-driven tools for chemistry. This is part of their initiative to shorten the drug discovery process and substantially reduce drug production costs.

Pfizeris another great example of AI powering pharmaceutical innovation during the COVID-19 pandemic. The company employed the help of Artificial Intelligence throughout its vaccine development process.

Most notably during its vaccine trials. Pfizer used its AI tech and data retrieval tools to parse through millions of data points to find signals in its 44,000-person COVID study. Even before the pandemic, Pfizer began digitizing its research and development operations.

It began a partnership with IBM’s Watson for drug discovery in 2016. Pfizer has also allied itself with Iktos, a virtual drug design firm, to leverage some of its Artificial Intelligence capabilities for improving drug discovery and production.

These examples are only the beginning. However, they provide enough evidence to indicate that the future of medicine and drug production will be heavily influenced by AI. This offers optimism for more effective and cheaper treatments and medication.

Other Ways Automation is Benefiting Healthcare

The risks and challenges of implementing AI and other automations in healthcare are far outweighed by the benefits. Not only can it reduce the cost of medications and treatments – it can also result in faster discovery of new medicines. Studies show that it leads to greater accuracy in imaging and preventing unnecessary biopsies.

5 Ways AI is Benefiting Healthcare:

  1. Early diagnosis – algorithms and machine learning speed up the processing of information and can help make accurate diagnosis.
  2. Cost reductions – automated process can access loads of data and lab results and make predictive results - which means less appointments for patients.
  3. Surgery assistance – AI surgical systems can now perform the smallest movements with complete accuracy – removing the risk of human error.
  4. Enhanced patient care – AI can automate scans of patient data and reports, and provide direction for patients and doctors.
  5. Improved information sharing – AI can track specific patient data, such as health monitors.

Automations in healthcare are not without challenges but each day AI implementation is helping to provide better and more affordable healthcare to larger and more diverse people groups.

Areas That Would Benefit Most From AI in HealthCare

Even with all the benefits of AI automations in healthcare, there are still many additional areas that could better benefit from AI. Technology costs for automation tools are becoming increasingly affordable – making the adoption of more AI tools increasingly feasible.

A 2020 study conducted by Statista’s official research department surveyed over 1,000 pharmaceutical and healthcare workers. The purpose of the study was to find which areas in healthcare would benefit from AI the most. 60% of respondents believed that AI had the potential to benefit quality control the most. Other areas included:

  • Customer care (44%)
  • Monitoring and diagnostics (42%)
  • Inventory management (31%)
  • Personalization of products and services (25%)
  • Cybersecurity (24%)

Once again, it’s important to understand the healthcare worker’s perspective. Artificial Intelligence has the potential to make jobs easier and more accurate. Understanding which areas healthcare workers believe AI is the most beneficial will be crucial to improving adoption.

Nevertheless, healthcare workers and employees are only a part of the equation. Directors, executives and leaders in the industry may be able to provide a more nuanced view. A2020 IDC report surveyed executives from 210 hospitals in the US(105), Germany (54) and the UK (51). They found that the use cases they were most concerned with were:

  • Inferencing to improve data quality (35%)
  • Reading images to assist in diagnosis (30%)
  • Early identification of hospital-acquired infection (30%)
  • Patient risk stratification (27%)
  • Improving back-office productivity (25%)
  • Predicting adverse events (25%)
  • Forecasting hospital patient admission (23%)
  • Supply chain (23%)
  • Early identification of sepsis (22%)
  • Chatbots for patient education and coaching (21%)

The Risks and Challenges of AI in Healthcare

Adoption of AI is not without concern and challenges. Arecent studyconducted by researchers at the Royal Free Hospital found that 80% of respondents (who were medical staff at the NHS foundation) had privacy concerns regarding the implementation of AI in healthcare.

This highlights one of the challenges of integrating technology and health care – the attitudes towards it. Some medical staff may be reluctant to use AI. Furthermore, it may require them to retool and relinquish the past approaches. However, some may argue that their fears are somewhat justified. AI often requires more robust network infrastructures, which may force hospitals to restructure their network security.

Data leaks or breaches may result in hefty fines for medical institutions. For instance, New York and Presbyterian Hospital (NYP) was forced to pay a $3.3 million settlement fee for HIPAA violations related to a data breach. To prevent situations like these, hospitals and other healthcare institutions will have to nurture a culture of cyber security savviness.

Even so, technology has never been perfect. It can be prone to bugs that may result in false positives/negatives and other issues. These issues may result in injury and death. When the failure of AI leads to dire consequences, who takes the blame? This failure creates unclear lines of accountability. The blame will most likely fall on the shoulders of the entire medical institution. Hence, many healthcare workers fear overreliance on Artificial Intelligence.

However, these are just some of the reasons why AI adoption has been slow. Others include:

Algorithmic and function limitations: While technology driving AI has advanced over the last ten years, it still has limitations. Most pundits worry most about the implications of issues such as bias in neural networks. Additionally, it may often be difficult for healthcare workers to trust or understand how AI produces results. There is very little transparency in the inner workings of AI.

Regulatory limitations and barriers: While regulations that focus on maintaining privacy are extremely important, they can stagnate development in AI. Again, lawmakers and those in charge of implementing AI for healthcare are most concerned about liability. Thus, it can take a long time for certain technologies to be approved, especially for an industry as sensitive as healthcare.

The Argument of ethics and incentives: Decision makers may retract from AI adoption because of the implications it may have on the medical profession. For instance, AI may replace certain tasks and thus lead to making certain jobs superfluous. Since decision-makers in healthcare tend to be practitioners themselves, they may not be too enthusiastic about replacing their peers and subordinates with AI.

Implications on Policy: In the long run, lawmakers will have to rewrite certain policies and laws to accommodate for AI’s eventual mass adoption in healthcare. Again, it is important to address liability. Of course, this will impact insurance companies and not just medical institutions.

Preparing for AI in Healthcare

Physicians and all medical staff members must first seek thorough training in the use of AI to curb the risks associated with AI in medicine. Furthermore, they must apply strict adherence to the rules and standards established by medical device and software companies.

Not only will this teach them the best practices related to using AI, but it will also give them the ability to articulate the potential risk to patients to obtain full informed consent.

The American Medical Association (AMA) has also suggested that AI training should be incorporated as a standard component of medical education. 

Additionally, hospitals and other medical practices are fundamental to ensuring the proper development, implementation and monitoring of the best practices and standards in the use of AI systems in healthcare. Healthcare workers can only benefit from AI related tools if they can understand how to use them. Technology that is too complicated or mis-understood will be under-utilized.

Directors and healthcare executives must take certain steps to ensure that the transition to and implementation of AI are smooth.

Steps to efficiently implementing AI

  1. Set a goal or machine learning statement for what you would like to achieve by implementing AI.
  2. Define data collection policies that consider ethical and demographic considerations.
  3. Research and focus on human-centric AI tools: this will make the technology less esoteric. If healthcare practitioners can understand how the technology fits into their daily routines, they’ll be more open to learning and working with those tools. Human-centric AI should feature:
    • Human override: the ability to override AI processes or tasks, so workers still feel that they are in control.
    • Human integration and not replacement: Workers should not feel that they are being replaced by AI. Thus, medical institutions should first assess tools that assist human workers. You should first aim for partial automation.

4. Train your workforce that will use the data. This should include:

  • Setting clear guidelines on the level of transparency required for training.
  • Understanding how data is gathered and recorded to make tools work.
  • Teaching the semantics and terms of AI technology.
  • Understanding where the AI originated.
  • The composition of AI tools - how they function, from the basics to some of the most complex processes.
  • The relationship between end-users, regulators and vendors.


AI is unlikely to completely replace medical professionals anytime soon. After all, the algorithms and technology that drive AI are just tools. And tools are utilized best when they are placed in the talented hands of professionals. Thus, physicians and medical staff who take advantage of the capabilities of AI will most likely replace those who do not – at least in the next 50 years.

We can expect AI in healthcare to make subtle but noticeable differences in the next few years. For instance, we’ll see more assistance with case triage, enhanced image scanning and segmentation, AI-supported diagnosis and improved workflows for medical professionals. Thanks to AI, the future of healthcare is encouraging.

Looking to automate your scheduling related tasks? See these six TimeTap features for healthcare facilities.

Learn how TimeTap can streamline your appointment booking tasks.

Book a Demo