Intro
One of the most impactful applications of AI is in healthcare. If leveraged responsibly, AI has the potential to positively transform the healthcare industry by improving patient care, reducing costs, and aiding drug development. However, use of AI in the healthcare industry also brings substantial risks. In this article, we’ll dive into some of the most common use cases of AI in healthcare to better understand the different risks and benefits to patients.
AI Governance in Healthcare
AI applications in healthcare are diverse. Use cases range from organization of health records, diagnostic medical imaging, aiding drug development, to disease prediction and prevention. In nearly all use cases, AI is positioned to radically transform healthcare delivery. With this transformation, however, comes substantial risks that must be considered for these AI algorithms to be considered responsible and ethical.
For example, Google Health is currently working with a handful of healthcare partners to test its ML system for mammography and has already published research that shows that their technology can identify signs of breast cancer as well or better than trained radiologists. Given these advancements, it might seem logical to supplement all manual mammography with google AI system, however, FDA clearance is required for medical devices before they can be used safely with patients.
Another area of interest is drug development. Given that the cost of developing prescription drugs has risen 145% over the last 15 years to approximately $2.6 Billion, companies are now looking towards AI and machine learning to repurpose existing drug trials. Successfully repurposing trial data, or data from retrospective cohort analyses, could change the way drug companies chose to develop new drugs or start new trials.
Still, there are other use cases with less acute sensitivity to false negatives, particularly in non-FDA cleared devices. For example, several digital health tech companies have developed AI powered wearable devices to enhance patient health monitoring. These platforms use machine learning algorithms to analyze data from sensors and electronic health records, enabling them to identify and anticipate issues like heart failure and strokes. Recently, there have been increased discussions about whether certain wearables should be classified as medical devices and subject to FDA clearance.
AI for Disease Identification and Diagnosis
Critical disease treatment is generally more effective the earlier it begins. Given that many diseases progress prior to the patient presenting symptoms, doctors are required to run labs, perform scans, and run other medical tests to determine if a patient has a certain pathogen. These labs and tests can be costly and time consuming, resulting in fewer being administered than are required to properly assess the population.
AI powered predictive models can serve a couple of different use cases here.
Firstly, AI can help identify areas of concern based on past medical history. These areas of concern can guide a physician towards ordering certain tests that they may not have previously considered but are appropriate for the patient. Depending on the robustness of the AI solution, new clinical programs and pathways for tests and labs can be developed, executed, and monitored. From these pathways, new standard clinical guidelines could eventually be developed based on AI algorithms, materially changing the way doctors diagnose their patients.
Secondly, AI can help in the review of tests and labs themselves, significantly reducing costs and saving time. It is already becoming more common for hospitals to integrate AI into their laboratory data workflow, seamlessly connecting lab results with other relevant patient information such as age, gender, and medical history for use within disease-specific predictive models. By combining this information, labs now can generate disease-specific patient probability while alerting physicians to potential patient risk diagnosis.
AI for Drug Discovery and Manufacturing
Drug discovery is notoriously time consuming and costly. It takes on average 3-6 years to clear a drug for over-the-counter use, and companies spend billions of dollars per year on this process. AI and Machine Learning tools are being used extensively in nearly every stage of the drug discovery process, helping to reduce costs and gain insights on potential drug efficacy before a trial is planned.
Recent developments in generative AI tools have the potential to transform drug discovery by enabling researchers to design new molecules and predict the efficacy of potential drug candidates through more accurate simulation.
Below are a few phases of drug discovery and development:
- Target identification
- Molecular simulations
- Prediction of drug properties
- De novo drug design
- Candidate drug prioritization
- Synthesis pathway generation
While each of these stages have used some form of AI over the past 10 years, computational biology and generics are very mature fields at this point, adding in the capabilities of generative AI models has brought a new dimension of potential advancements and risks to the field.
Clinical Documentation
By utilizing machine learning and natural language processing, AI can transcribe conversations between patients and providers in real time, with the goal of freeing clinicians to focus more on patient care. Amazon Web Services recently announced their new AI-powered service for healthcare software providers that will help clinicians with paperwork. The service can create transcripts, extract key details, such as medical terms and medications, and create summaries from doctor-patient discussions that can then be entered intoan electronic health record (EHR) system.
Clinical AI and Predictive Analytics
Boston-based healthcare technology company Health Data Analytics institute has partnered with Houston Methodist hospital to deploy clinical AI at scale. According to a recent press release, HDAI’s HealthVision™ platform has been implemented system-wide and is “one of the most comprehensive AI/ML care optimization solutions in the world as well as one of the largest deployments of generative AI within healthcare today.” HDAI’s uses AI to process over hundreds of thousands of medical records each month, ensuring that care-teams have the most up-to-date predictions and insights.
These insights can be used to help clinicians adjust for risk and better understand their patient populations. This in-turn can help them deliver better and more cost-effective care, while reducing physician burnout.
Deep Learning for Healthcare
There are many deep learning concepts being applied in science and medicine. According to this article, MD Anderson, data scientists have developed a deep learning algorithm to predict acute toxicities inpatients receiving radiation therapy for head and neck cancers. In clinical workflows, the medical data generated by deep learning algorithms can identify complex patterns automatically and offer a primary care provider clinical decision support at the point of care within the electronic health record.
Gigantic amounts of unstructured healthcare data represent almost 80% of the information held or “locked” in electronic health record systems. Machine and Deep learning can now mine that data. These are relevant data documents or text files with patient information, which in the past could not be analyzed by healthcare machine learning but required a human to read through the medical records
AI for Medical Imaging
AI analysis for medical imaging helps healthcare professionals identify problem areas or details that may be missed by the human eye. For example, AI-powered medical imaging can analyze data points in a medical report to distinguish a disease. AI-based medical imaging is being used to:
- Identify complex patterns in imaging data
- Provide a quantitative evaluation of radiographic traits.
- Detect image modalities at various treatment stages (for example, tumor delineation).
- Discover various disease characteristics that are not detectable by human eyes.
AI and Rehabilitation
Several activities related to rehabilitation could potentially take advantage of artificial intelligence techniques. Some of these include:
- Examination of physical and cognitive behavior.
- Continuous measurement (Remotely as well).
- Risk estimation via measurement data.
- Physical (robotics, exoskeleton) and Virtual (VR, AR).
- Streamlining and automating admin tasks.
AI for Hospital Logistics
Healthcare companies are implementing AI to streamline workflows, optimize routes, forecast demand, and enhance real-time tracking. AI can help deliver sensitive cargo in a timely and secure fashion while enabling healthcare professionals to focus on patients. Examples include:
- Smart Warehouses with Robotics
- Touch-less Deliveries of Patient Samples
- Demand Forecasting and End-to-End Transparency
- More Efficient Fleet Management
- Sustainable Operations and Optimized Routes
AI and Dosing Recommendations
For medication management, AI can cross-verify patient data, prescription data, and pharmacy inventory. It can identify potential issues such as incorrect dosages, drug-drug interactions, drug-disease contraindications, and duplicate therapies.
AI can provide real-time guidance to pharmacists during medication prescribing, dispensing, and administration processes, flagging duplicate prescriptions, potential drug interactions, allergies, or contraindications, reducing the likelihood of errors.
Medication reminder apps can also aid patients in adhering to their medication regimen, improving outcomes, and reducing hospital visits.
Chatbots and Virtual Health Assistants
AI-powered healthcare virtual assistants can deftly handle simple tasks so that skilled medical professionals can use their time to effectively manage more complex jobs that they have been trained for.
AI-powered virtual assistants in the healthcare sector have gained prominence as health bots continue to provide benefits such as:
- Reduced waiting times
- Massive reduction in care costs
- Timely medical advice
- Free up doctors
- Anonymity
- Real-time interaction
- Scalability
- Patient satisfaction
AI Symptom Checker
There are many online AI powered symptom checkers on the market and a wide variation in performance. Overall performance is significantly below what would be accepted in any other medical field, though some do achieve a good level of accuracy and safety of disposition. External validation and regulation are urgently required to ensure these public facing tools are safe.
AI for Mental Health Assistance
Chatbots are increasingly being used to offer advice and as a line of communication for mental health patients during their treatment. They can help patients cope with symptoms and monitor for trigger words which would lead to referral and direct contact with a human mental health professional.A diagnostic tool from Limbic, an AI startup, has screened more than 210,000 patients and are boasting a 93% accuracy across the most common mental disorders, including anxiety, depression, and PTSD.
AI Wearables for Mental Health
Some AI mental health solutions function as wearables that can interpret signals using sensors and step in to offer help when it's needed. A product called Biobeat collects information on sleeping patterns, physical activity, and variations in heart rate and rhythm that are used to assess the user’s mood and cognitive state. This data is compared with aggregated and anonymized data from other users to provide predictive warnings when intervention may be necessary.
AI and Patient Triage
A team of researchers from Yale University and other institutions globally has developed an innovative patient triage platform powered by artificial intelligence (AI) that the researchers say is capable of predicting patient disease severity and length of hospitalization during a viral outbreak.
The platform, which leverages machine learning and metabolomics data, is intended to improve patient management, and help health care providers allocate resources more efficiently during severe viral outbreaks that can quickly overwhelm local health care systems. Metabolomics is the study of small molecules related to cell metabolism.
Patient Monitoring with AI
AI is advancing the field of Remote Patient Monitoring (RPM), improving efficiency, and enabling early intervention. RPM leverages technology to monitor patients’ health conditions remotely, detecting trends, anomalies, and potential issues. This may reduce the need for frequent in-person visits.
AI Patient data for AI Telemedicine
Teledoc Health, the largest telehealth company is tapping Microsoft to integrate AI and clinical documentation tech into its Solo virtual care platform, the two companies announced recently.
Some of the Key Components:
- Real-Time Monitoring: Wearable devices and sensors
- Pattern Recognition
- Anomaly Detection
- Predictive Analysis
AI Fraud Detection in Healthcare
Fraud detection in medical claims has been historically time-consuming and costly. ML models are being used to automate claims assessment and routing based on existing data related to fraud patterns. This process flags potentially fraudulent claims for further review before approval and payment. More advanced anomaly detection systems can be deployed to find new patterns and to flag those for review, which leads to prompt investigation of new fraud types.AI systems can also provide clear reason codes, so investigators can quickly see the key factors that led the AI to flag for fraud to begin with. With AI powered fraud detection tools, fraudulent claims can be flagged before they are paid.
Summary
Given the vast amount of data that is collected in the healthcare industry numerous AI solutions are available on the market. These solutions have the potential to transform and improve the way patients receive care. They also bring substantial risks if not properly governed. Fairo can help your organization continue its pace rapid innovation while maintaining ethical and regulatory standards around your AI systems.
How Can Fairo Help?
Fairo is a SaaS platform focused on standards, simplicity, and governance to give organizations and their users the confidence to consume AI successfully and rapidly at scale. Fairo is committed to being the industry-standard platform for helping yourorganization implement its AI governance framework and strategy. Fairo seamlessly integrates into your existing ecosystem and is easy to consume. AI is a disruptive technology that will change how people work and live. We envision a world where AI is universally built responsibly, trusted, and not feared. We aim to provide an easy-to-use solution that helps organizations procure, develop, and deploy trustworthy AI solutions with confidence.
