It seems as though everyone is talking about artificial intelligence, usually referred to as AI, these days! Indeed, not only are AI tools now accessible to everyone, but many of us are already using AI regularly, sometimes without realizing it via social media, online shopping, smart home devices, or streaming platforms.
AI is the use of computer technology to learn from experience and analyze new and large amounts of information to perform tasks generally associated with human intelligence and is an umbrella term that includes machine learning, natural language processing and deep learning [1]. It can include data analysis, image recognition, translation, text and image generation, problem solving, and decision-making [2, 3], and while AI is already used across many industries, including financial, e-commerce, manufacturing, and healthcare, it’s use is likely to continue to expand [4].
In this article, we look at the role of AI in healthcare, including how AI can take on administrative tasks, aid physicians with diagnosis and treatment plans, assist researchers in their quest to identify potential drugs, and support patients in taking charge of their healthcare journey. Some of the points covered are already in everyday use, such as medical chatbots, while others represent areas in development or the potential of AI technology.
Contents
The role of AI in healthcare administration
Administrative tasks
The ability of AI to process large amounts of data makes it ideal for streamlining administrative tasks, such as billing and claim processing [5], documentation and data management [6]. It can also streamline scheduling and confirming appointments [5]. This can reduce the time healthcare providers require for such tasks and increase both the time available for patient contact and the speed with which they can see patients [7].
Enhancing clinical operations
The use of AI technologies can enhance clinical operations, for example by optimizing staff schedules and improving administrative functions [8, 9], or by ensuring stock of essential medical supplies [10]. It can also make projections based on past trends further increasing the efficiency of operations [7].
Recruitment
AI can review CVs and shortlist candidates for interview, thus decreasing the time required to fill open healthcare positions and ultimately further enhancing clinical operations [9].
Data security
AI has a role in increasing cybersecurity and this may become increasingly relevant for the protection of sensitive health data [5].
The role of AI in medical diagnosis and treatment
Image processing
Machine learning algorithms can learn to analyze medical images (e.g., X-rays, CT and MRI scans, and ultrasound), and detect abnormalities, positive or negative results, or changes that are too small for humans to easily detect [3, 11, 12]. It can process a large amount of data very quickly, enabling clinicians to provide more accurate and faster diagnoses, which can improve outcomes [11, 12] and may positively affect referral and patient waiting times [13]. Examples include detecting breast cancer on mammograms, lung cancer on CT scans or Alzheimer’s disease on MRI scans [12]. In fact, AI has been shown to be at least as good as experts in diagnosis using images in several fields, including dermatology, pathology, cardiology, and detecting pneumonia [14]. AI can also aid risk assessment and identifying risk factors for certain diseases [12]. AI based screening procedures may also be more consistent than traditional methods, as they are not affected by human factors such as tiredness or distraction [13].
Use in pathology
Pathology is one field where AI is being increasingly used. In 2017, the FDA approved the use of the first automated whole slide imaging scanner for the interpretation of digital surgical pathology slides prepared from biopsied tissue, commenting also that it has the potential to streamline slide storage and retrieval [15]. Furthermore, 4 years later in 2021, the FDA approved an AI-based software for the analysis of prostate biopsies for the detection of prostate cancer, which during trials led to improved prostate cancer detection rates [16].
Automating medical imaging tasks
AI can automate medical imaging tasks such as processing images, quality control, labelling images, and data management. This can provide healthcare providers with a complete patient record, and allow them to work more efficiently, improving the accuracy of diagnoses [12]
Electronic health records
AI can not only compile electronic health records [10], it quickly and efficiently analyze a patient’s complete electronic health record to identify risk factors, make diagnoses, and predict disease progression [11, 12, 14] and convert complex medical charts into easy to understand reports [11, 12].
Providing treatment recommendations
One exciting application of AI is how it can assist healthcare providers by providing diagnostic recommendations based on patient symptoms and data, including by reviewing a large amount of healthcare data from different sources, including electronic health records [11, 14]. Incorporating patient specific information such as genetics, medical history and lifestyle enables AI to determine more personalized treatment plans [12]. It can also use mathematical modeling and statistical analysis to predict treatment outcomes [11], benefiting the patient by trying the treatment most to work and improve patient outcomes, as well as saving both time and money.
Robotics
Robotic surgery involves a surgeon controlling the robot from a computer. AI could improve the accuracy of robotic surgery [11] and robot-assisted surgery is associated with better treatment outcomes [1].
The role of AI in improving patient care
Chatbots and virtual assistants
Patients can already use AI-powered virtual assistants and chatbots in some fields to check symptoms and schedule appointments [14], suggest follow-up actions, remind about appointments [1] or answer other patient questions [5, 10], which has the potential to increase staff efficiency and patient satisfaction.
Wearable devices
People are taking active responsibility for their own health and this includes wearable devices such as smartwatches that provide the user health and wellness data in an instant. These devices use basic AI powered algorithms to provide the wearer with useable health metric information. It can also provide user recommendations based on inputted health goals. Furthermore, AI algorithms can assess notifications and battery usage and may power voice assistants [17].
Streamlined administrative and monitoring processes
Streamlined processes within healthcare settings and quicker diagnosis driven by AI, have the potential to lead to increased patient satisfaction and improved treatment outcomes. Automating routine tasks has the potential to increase the time available for patient contact.
AI-driven sensors could assist patient monitoring and provide timely warning of adverse events, deterioration or even hospital re-admission [7, 14].
Patient benefits and personalized medicine
As well as potentially increasing the speed of diagnoses, AI offers the chance for earlier disease detection, consistent analysis of data [13] disease risk assessment and prognosis [12]. These previously discussed advantages of using AI as well as the combined analysis of previous medical history and genetic information illustrate the potential to provide personalized treatment plans tailored to a patient’s situation and lifestyle. Such plans may include optimized medication and dosage information, which has the potential to offer better treatment outcomes. Assessing a patient’s individual health profile may lead to disease prevention or earlier diagnosis [18]. Digital twins could also have a role in assessing the success of treatment options in a personalized approach, particularly in oncology [19].
AI can use predictive modeling to analyze a large amount of information and predict how patients will respond to certain treatments or likely outcomes based on patient-specific information [12].
Compliance
AI’s ability to scan huge amounts of data means it can identify areas or non-compliance to protocols, standards or regulations. This could reduce the risk of adverse events for patients and improve outcomes [10].
The role of AI in drug discovery and development
Research and development for diagnosis
AI can assess data from different sources and interpret how it can be combined to benefit patient diagnosis. This can also assist in the development of new diagnostic tools [12].
Drug discovery and clinical trials
With access to a vast and ever increasing amount of biological data and the ability to analyze it quickly and efficiently, AI has the potential to optimize the drug discovery process, by identifying molecules likely to be effective or designing drugs and predicting adverse side effects [1, 14]. It can also identify new drug targets, combinations, or ways in which existing drugs can be repurposed [1]. This will increase the efficiency and reduce the speed and cost of the development process. Potential drugs can be tested through AI designed and optimized clinical trials and AI can identify ideal test populations [1, 14].
Digital twins are virtual AI-driven patient data models that can replicate body functions, organs or body systems, or the whole body in health or disease states. They are being used to accelerate and streamline drug development by identifying biomarkers or drug targets that are likely to be successful, as well as to assess the safety and efficacy of new drugs before they are tested in patients in clinical trials. This can increase clinical trial efficiency. Digital twins can also be used for medical device design, surgical planning, clinical trial planning, and personalized medicine [19, 20]. However, digital twins are not without drawbacks, not least the massive amounts of patient data required for effective use and the technological challenges involved in managing this data [19].
The role of AI in training and education
AI has a growing role in training and education in the medical profession. For example, pathologists can already preview digital slides and use annotation tools to create an interactive learning experience [21].
Ethical and regulatory considerations and other challenges
Despite the considerable potential of AI in healthcare and diagnostics, its use is still limited and progress is relatively slow. Some reasons for this include:
Privacy and data security concerns: There needs to be a focus on patient safety and confidentiality. Machine learning requires access to a large quantity of good quality data, while recognizing the sensitive nature of patient data and privacy concerns [12, 14]. This may require healthcare organizations to review how their data is organized and how they protect access to the data [7, 10]. Data needs to include not only the target patient population, but also other conditions, patients and environments to avoid unintentional bias [13].
Transparency: Transparency in data protection and how AI algorithms reach important decisions is essential [9].
Functionality and usability: It is necessary to demonstrate capabilities in real-life clinical environments and controlled studies that include publications in reputable journals, and healthcare providers may be reluctant to use new technologies without this [13]. Furthermore, AI and machine-learning applications need to meet actual clinical needs and be useable in real-life settings, within existing workflows [13] and compatible with existing systems [7].
Regulatory concerns: There is a need for guidance regarding the development and use of AI in healthcare and diagnostics, in particular for regulatory requirements and obtaining regulatory approvals based on safety and effectiveness for AI software [11, 13]. Local and national monitoring centers have been suggested as a means to monitor AI performance and safety [11, 14].
Bias: It is necessary to ensure that AI applications are not only effective and safe for patients, but that there is no bias towards specific groups [7].
Liability: There should be clear guidelines for the use of AI and the chain of accountability and liability should be clear in case of errors. It has been recommended that AI be considered as a tool to support healthcare providers rather than as a replacement and that experts retain the autonomy to over-rule AI-based recommendations [7, 11, 12, 14, 18].
IT hardware and security: There is a need for computers and IT solutions that are sufficiently powerful to successfully support healthcare providers using AI [14] and that are well protected from cyber threats [7, 11].
Training: Clinicians may require extensive theoretical and practical training before they are able to use AI systems. This needs to include an understanding of potentially complex algorithms and their potential limitations, as well as liability issues and the importance of human judgement [9, 11].
Costs: There may be significant ongoing costs involved in terms of software, hardware, consultants, and training [10].
Informed consent: It should be clear to patients when AI is used in their care and they should be informed about the risks and limitations [7].
Fear and reluctance: Fear around job losses due to routine task automation or the use of chatbots [9]. There may also be patient reluctance to use chatbots [6].
Future Trends
Much of the information presented in this article represents the future possibilities of AI. Once any challenges have been overcome, AI has the potential to drive fully comprehensive, connected healthcare through the complete integration of information provided by wearable devices, sensors and other connected medical devices, with electronic health records, genetic information and myriad other health data to provide fully comprehensive healthcare that is data-driven, preventative and personalized [7, 14].
Conclusion
The use of AI holds immense potential for improving healthcare and patient outcomes, from enabling healthcare providers to spend more time with patients and make quicker diagnosis and treatment decisions, to accelerating new treatment options [6] and providing more personalized patient care [12, 19]. Furthermore, it offers a chance for facilities like hospitals to control costs by reducing readmissions due to missed/delayed diagnoses and optimizing staff resources [8]. While image analysis by AI and machine learning is in use for certain diseases including some cancers, diabetic retinopathy, Alzheimer’s disease and heart disease, its use is still limited [13] and key challenges around ethics, safety, data privacy, regulation, and liability remain [11, 13],
However, if these concerns are addressed to balance them with inevitable technological advances, it is clear that AI has enormous potential to streamline, shape and even revolutionize healthcare.
AI Disclaimer: This outline for this article was written with the help of AI.
References
[1] Daley, Sam. “AI in Healthcare: Uses, Examples and Benefits.” Built In, 24 Mar. 2023, https://builtin.com/artificial-intelligence/artificial-intelligence-healthcare. Accessed 19 Aug. 2024.
[2] Emeritus. “Examples of Artificial Intelligence (AI) in 7 Industries.” Emeritus – Online Certificate Courses | Diploma Programs, 4 Mar. 2022, https://emeritus.org/blog/examples-of-artificial-intelligence-ai/ Accessed 10 Jun. 2024.
[3] SAS. “Artificial Intelligence – What It Is and Why It Matters.” SAS, 2023, www.sas.com/en_us/insights/analytics/what-is-artificial-intelligence.html. Accessed 10 Jun. 2024.
[4] Forbes. “Applications of Artificial Intelligence across Various Industries.” Forbes, 6 Jan. 2023, https://www.forbes.com/sites/qai/2023/01/06/applications-of-artificial-intelligence/. Accessed 10 Jun. 2024.
[5] Lee, Yu. “The AI Revolution in Healthcare Administration | NCC.” Northwest Career College, 12 Mar. 2024, www.northwestcareercollege.edu/blog5/the-ai-revolution-in-healthcare-administration/. Accessed 19 Aug. 24.
[6] Davenport, Thomas, and Ravi Kalakota. “The potential for artificial intelligence in healthcare.” Future healthcare journal vol. 6,2 (2019): 94-98. doi:10.7861/futurehosp.6-2-94 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6616181/
[7] Bhagat, Shefali V, and Deepika Kanyal. “Navigating the Future: The Transformative Impact of Artificial Intelligence on Hospital Management- A Comprehensive Review.” Cureus vol. 16,2 e54518. 20 Feb. 2024, doi:10.7759/cureus.54518, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10955674/.
[8] SAS “Artificial Intelligence (AI) Solutions for Health Care.” www.sas.com, www.sas.com/en_us/industry/health-care/technology/ai.html. Accessed 29 Jul. 2024.
[9]. “The Profound Impact of AI in Healthcare Administration | Shiftmed Blog.” www.shiftmed.com, 15 Feb. 2024, www.shiftmed.com/blog/impact-of-ai-in-healthcare-administration/. Accessed 21 Aug. 2024.
[10] Puri, Gauri. “Healthcare Reimagined: Harnessing AI to Transform Administration | Viewpoint.” OncLive, 21 Jan. 2024, www.chiefhealthcareexecutive.com/view/healthcare-reimagined-harnessing-ai-to-transform-administration-viewpoint. Accessed 21 Aug. 2024.
[11] Park, Chan Woo et al. “Artificial Intelligence in Health Care: Current Applications and Issues.” Journal of Korean medical science vol. 35,42 e379. 2 Nov. 2020, doi:10.3346/jkms.2020.35.e379 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7606883/
[12] “Revolutionising Medical Imaging with AI and Big Data Analytics.” Open Medscience, 18 Mar. 2023, https://openmedscience.com/revolutionising-medical-imaging-with-ai-and-big-data-analytics/. Accessed 29 Jul. 2024.
[13] U. S. Government Accountability Office. “Artificial Intelligence in Health Care: Benefits and Challenges of Machine Learning Technologies for Medical Diagnostics.” www.gao.gov, 29 Sept. 2022, www.gao.gov/products/gao-22-104629. https://www.gao.gov/assets/730/723160.pdf. Accessed 21 Aug. 2024.
[14] Bajwa, Junaid et al. “Artificial intelligence in healthcare: transforming the practice of medicine.” Future healthcare journal vol. 8,2 (2021): e188-e194. doi:10.7861/fhj.2021-0095, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8285156/.
[15] Office of the Commissioner. “FDA Allows Marketing of First Whole Slide Imaging System for Digital Pathology.” U.S. Food and Drug Administration, 2019, www.fda.gov/news-events/press-announcements/fda-allows-marketing-first-whole-slide-imaging-system-digital-pathology. Accessed 29 Jul. 2024.
[16] Commissioner, Office of the. “FDA Authorizes Software That Can Help Identify Prostate Cancer.” FDA, 01 Oct. 2021, www.fda.gov/news-events/press-announcements/fda-authorizes-software-can-help-identify-prostate-cancer. Accessed 29 Jul. 2024.
[17] Waseem, M. “Ways AI Is Enhancing Smart Watches for Everyday Living.” Medium, Medium, 12 Mar. 2024, https://medium.com/@mwaseem85526/ways-ai-is-enhancing-smart-watches-for-everyday-living-87bb8ee0f493. Accessed 19 Aug. 2024.
[18] Schaar, Mihaela van der. “AI-Powered Personalised Medicine Could Revolutionise Healthcare (and No, We’re Not Putting ChatGPT in Charge).” The Guardian, 26 June 2023, www.theguardian.com/commentisfree/2023/jun/26/ai-personalise-medicine-patient-lab-health-diagnosis-cambridge. Accessed 19 Aug. 2024.
[19] Katsoulakis, E., Wang, Q., Wu, H. et al. “Digital twins for health: a scoping review.” npj Digit. Med, 7, 77 (2024). https://doi.org/10.1038/s41746-024-01073-0, https://www.nature.com/articles/s41746-024-01073-0, https://www.nature.com/articles/s41746-024-01073-0.pdf.
[20] “Digital Twins in Clinical Trials | Healthcare Trends 2024.” www.definitivehc.com, www.definitivehc.com/blog/digital-twins-clinical-trials. Accessed 19 Aug. 2024.
[21] Shafi, Saba, and Anil V Parwani. “Artificial intelligence in diagnostic pathology.” Diagnostic pathology vol. 18,1 109. 3 Oct. 2023, doi:10.1186/s13000-023-01375-z, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10546747/, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10546747/pdf/13000_2023_Article_1375.pdf
