Senior Innovation Manager,
Executive Director of Innovation and Research
Chief Innovation Officer,
It’s hard to talk about the future of healthcare (or any industry for that matter) without discussing the role of Artificial Intelligence (AI). AI, and machine leaning, represent a significant opportunity to revolutionize the global healthcare ecosystem. Click here for AI and machine learning definitions from the FDA. As both computing power and data storage capabilities continue to increase, the opportunity for machine learning to impact physician decision making, personalized medicine, and enhance the quality of patient care increases too.
There is ample opportunity surrounding automation of business operations, logistics, and administrative processes to greatly reduce red tape for patients and administrators throughout the health system. For physicians, AI and machine learning providevaluable tools to help navigate what sometimes seems like unmanageable troves of data. AI and machine learning can, help lessen physician burn out, too. As Eric Topol has argued in his latest book “Deep Medicine,” -artificial intelligence has the potential to make healthcare human again.
Current Applications at LifeBridge Health
While many of these advances are years away from realization, LifeBridge Health has been actively using AI applications since 2016, resulting in a number of successful machine learning applications. Both the radiology and ophthalmology departments have committed to piloting with companies that are implementing AI and machine learning solutions with existing imaging and diagnostic capabilities. Here are three examples of how LifeBridge is using AI today:
|1. iSchemaView’s RAPID technology – An imaging platform that applies deep learning algorithms to CT scans to help locate parts of the brain that are not currently receiving enough blood flow, which allows for faster triage and intervention. Where traditional diagnostic processes can take a physician an hour to complete, RAPID provides automated analysis in as little as 90 seconds. Since implementation in 2016, LifeBridge has quadrupled the number of life-saving thrombectomy procedures offered by the hospital to the Baltimore community.|
|2. Intelligent Retinal Imaging Systems (IRIS)– IRIS is a leader in early detection systems for diabetic retinopathy and other retinal pathologies. Their end-to-end diagnostic solution is designed to work within a primary care setting and helps to increase access and create better outcomes for patients with sight-threatening disease. IRIS applies machine learning by leveraging the largest database of retinal images with 95% accuracy – surpassing the accuracy of human readers. Since implementation, IRIS has made approximately 100 saves, preventing vision loss or blindness.|
|3. Aidoc Medical – Aidoc helps radiologists rapidly identify patients that require the most immediate care with its imaging AI software. Through deep learning and AI algorithms that analyze medical images and patient data, their product helps detect and pinpoint critical anomalies for radiologists. Aidoc eases up the provider work list and frees up time and attention to what matters. For instance, LifeBridge Health has achieved EMR integration with Aidoc’s software analyzing Head CT and Abdomen CT studies at our Northwest and Sinai hospital locations to flag intracranial hemorrhage and pulmonary embolism. We are in the process of implementing a workflow that combines the strengths and speed of the algorithm alongside our clinical staff to achieve improved accuracy and efficiency. We anticipate by years end we will be using this “human in the loop” imaging AI workflow to triage scans of over 100 emergency patients per day including all trauma patients.|
Balancing the Risks and Rewards
As with any new emerging technology, AI presents a new and unprecedented set of opportunities and challenges. The potential to reduce spend and improve patient outcomes and access to care with AI solutions is not without inherent risks. These include ethical issues, privacy concerns, and potential for medical error. While it’s certain that human involvement will remain a necessary and valuable component of AI workflows for the foreseeable future, it’s unlikely that we will see a profusion of medical error as a result. The more likely outcome will be increasing false positives, which creates a new set of burdens with potential impact on ROI. However, this is not necessarily a new problem with implementation of novel diagnostic technology, and one that we have to overcome multiple times in the past in a healthcare setting.
While diagnostic applications are promising, the operational and administrative use cases are equally, if not more compelling in the near term. As high reliability organizations, hospital operations are highly complex and risk averse, and simultaneously deal with large volumes of protected data. In the current paradigm, health care systems struggle to function dynamically, because care teams are using static and siloed data. In a future paradigm that leverages the right AI tools, front line staff can get access to real-time data to help better inform decision-making.
Despite the remaining barriers to widespread adoption of AI and machine learning, public and private sector investors are not deterred. According to CB Insights, AI healthcare start-ups have raised $4.3 billion since 2013, which exceeds the AI activity in all other industries.
The uptick in investment dollars might, in part, be explained by the expected value of the potential annual benefits of AI and machine learning technology. Some strategists forecast that by 2026, the current AI market will grow by more than 10x, and that the top 10 AI application will offer savings to the U.S. healthcare systems of upwards of $150 billion.