Discover how Aidoc’s AI platform offers seamless end-to-end integration into a facility’s existing IT infrastructure enabling implementation of AI at scale.
Discover how Aidoc’s AI platform offers seamless end-to-end integration into a facility’s existing IT infrastructure enabling implementation of AI at scale.
Innovators, engineers, and entrepreneurs are attracted to the goal of creating disruptions in healthcare for 3 main reasons: 1) improving the lives of others is phenomenally meaningful, 2) the overarching problems facing healthcare are one of the most complex and multifaceted challenges facing humanity, and 3) there is excitement in working as multidisciplinary teams towards the goal of improving the health of others. Whether it be engineering new brain-computer interfaces or working on economic solutions to making healthcare more affordable, the list of areas within healthcare that are ripe for innovation is endless. To that end, nearly every subdomain within medicine will eventually be impacted by the integration of artificial intelligence.
From genomic sequencing to pharmaceutical pricing, AI offers unprecedented insights into both the human body and the healthcare ecosystem. The acceptance of AI within the clinical setting catalyzed its capabilities as it pertains to coordinating care for patients. Ensuring that each patient entering a hospital receives the right care, by the right physician, at the right time, has fallen into the purview of AI capabilities. Working at the intersection of big health data and radiology, AI for care coordination offers health systems and physicians an opportunity to treat each patient seeking care with the most appropriate clinical intervention.
Understanding patient care coordination challenges
As patients navigate the complexities of the healthcare system, their encounters with healthcare providers generate nodes of data. A patient with congestive heart failure may initially visit their primary care provider, obtain blood work and lab values, and subsequently meet with their cardiologist and undergo extensive cardiac imaging. Each of these interactions creates a wealth of health information that, when synchronized and shared appropriately, may improve patient outcomes and decrease medical costs. However, without the ability to seamlessly share and integrate data points from every patient encounter, healthcare data becomes siloed and negligible. It is only when healthcare data can be translated into actionable insights that it is of value to the overall coordination of care for the patient. Consequently, the core of care coordination is healthcare interoperability.
Interoperability in medicine is simply the capability of different technological systems to share information and data without extensive effort on the part of the user. With regulatory pressures increasing from the Centers for Medicare and Medicaid Services (CMS), the U.S. healthcare system as a whole is making progress in progressively augmenting interoperability throughout the ecosystem. The Agency for Healthcare Research and Quality defines care coordination as “the deliberate organization of patient care activities between two or more participants (including the patient) involved in a patient’s care to facilitate the appropriate delivery of health care services. Organizing care involves the marshalling of personnel and other resources needed to carry out all required patient care activities, and is often managed by the exchange of information among participants responsible for different aspects of care.” When referring to care coordination and the power of interdisciplinary communication of data, we are highlighting two components within the administration of medical care: a) care provided to a patient over the span of years, evolving throughout a patient’s illness or management of a chronic disease, or b) the coordination of care within a given hospital admission, whereby efforts are focused on delivering the right care, to the right patient, at the right time. Both care coordination models, though, are hindered by common obstacles: quality data, complete data, and the interoperability of that data.
Barriers in the hospital care coordination model
The CDC reports that data from the 2018 National Health Interview Survey (NHIS) demonstrated that 51.8% of adults had at least 1 of 10 selected diagnosed chronic conditions, and 27.2% of American adults had more than one chronic medical condition. As the chronically ill population faces a more fragmented healthcare system, care coordination programs have become increasingly necessary. The benefits of care coordination to the patient and the healthcare system involve both increased quality and decreased cost; specifically, innovation in the care coordination space may lead to reduced hospital admissions, more efficacious interventions for the management of chronic disease states, easier access to medical specialists and improved patient satisfaction. Research and development by both industry and academia continues to sprout forth novel care coordination models that directly alleviate the struggles physicians face in administering care.
From a physician’s perspective, a hospital care coordination model must be designed to assess the clinical needs of patients and assist in leading them through the ecosystem with the intent of managing their underlying health concerns. The structure of the healthcare system itself propagates the administration of care in a fashion that is siloed and secluded from the interdisciplinary nature by which medicine must be practiced. When cardiologists and primary care physicians are unable to sequentially communicate or seamlessly share patient data, this results in fragmented care that frustrates physicians, burdens the overall healthcare system, and leads to poor patient outcomes. Inadequate information exchanges amongst a diverse set of specialists creates a lack of care coordination that may direct a patient to the inappropriate place in the healthcare system. By enabling physicians to work together in an integrated fashion through communicative platforms, the isolated service lines of medicine are dismantled, and patients can receive the best care the system may offer.
An important effect of solving barriers to hospital care coordination is the wellness of physicians. The modern healthcare system in the United States has catalyzed an ever-growing percentage of physicians reporting burnout. Studies by the Mayo Clinic report that over 40% of physicians reported feelings of burnout over the last several years, with an increase noted during the Covid-19 pandemic. Technological advancements that aim to innovate in the healthcare sector must focus on decreasing burnout among physicians; it is a prerequisite for products, technologies and solutions that enter the healthcare ecosystem to mitigate the frustrations felt by nurses and physicians alike. To that end, care coordination models that are focused on delegating care in a succinct manner may aid in reducing physician burnout. Specifically, physicians may be contacted for a variety of cases and interventions that do not necessitate their input. When care is uncoordinated, consultations to the wrong service line can disrupt physicians that are not necessarily on-call, leading to fatigue and burnout. With the advent of care coordination systems within a hospital or network of care, doctors are consulted strategically and precisely when their services are most appropriate. The byproduct of hospital care coordination isn’t limited to reducing physician burnout. Through building technologies that facilitate coordinated input and communication from a variety of sub-specialties, a hospital inevitably improves its ability to allocate the right medical services to the right patient at the right time. This ensures that the right surgeon or most adept service line is consulted for the precise needs of the presenting patient.
The role of AI in data management
Advanced algorithmic power and increased computer processing capabilities ushered in an era in which big data is the ultimate currency. From operational management to logistics and healthcare, the accumulation of large sets of data unleashes insights into processes, plans and people that organizations have yet to discover hitherto. According to a Dell Study, the average enterprise is now managing 10x more data than it did five years ago, from 1.45 petabytes to 14.6 petabytes. Within all arenas of life, it is becoming increasingly clear that data is invaluable in the decision-making process. Perhaps, though, there is a limit to which data is found useful? Too much data could be harmful if it overwhelms one’s ability to think fluidly and clearly about a problem, and disorganized data may wreak havoc on an organization’s ability to leverage big data into actionable insights on a pragmatic level. For big data to be useful, it needs to be clean, organized, succinct, and correct. We look at data as a blueprint, as a representation of what reality may currently look like, and how it could morph into something better in the future.
Data within healthcare has gained phenomenal acceptance as a tool directly responsible for better patient outcomes and cost containment. Unlike other sectors that leverage data for decision-making processes, big data in healthcare comes from a variety of diverse sets of data that may be synthesized together towards the goal of improving healthcare administration. Electronic health records, genomic studies, radiologic imaging, claims and payor data, medical devices and lab studies are but a few sources that join together to form oceans of data within the healthcare ecosystem. Due to the nature of where health data comes from, managing its volume and content is a major obstacle hospitals and health systems face. Furthermore, what is strikingly different about health data is that it aggregates at extraordinarily high volume while spanning the entire industry’s digital universe at a very high speed. It isn’t only the volume of healthcare data that makes it challenging to manage, but also its heterogeneous structure, format and inherent nature that often prevents healthcare providers and systems from digging deeper into large data oceans.
The framework for harnessing the power of AI in managing large sets of health data is best understood by dividing AI decision-making processes into three general categories. As put forth by a study from Gartner, each category offers its own unique advantage towards arriving at the best decision possible, given the available data and the complexity of the issue at hand:
Decision automation: In this framework, the AI system manages large sets of data lakes in an effort to arrive at meaningful conclusions. The AI system itself is making decisions based on prescriptive or predictive analytics. Within the clinical world, this might be an AI-system creating a call schedule in an emergency department. Using AI in the management of data in this manner is the fastest and usually offers the most consistency in decision making.
Decision augmentation: While also using predictive analytics, an AI-based system may recommend that a physician take certain actions towards the benefit of a patient. A clinical scenario may involve an AI-algorithm recommending placing a patient on a statin based on their cholesterol levels, family history, EKG data, recent visits to the emergency department, and current medications used. A data team may mine through billions of nodes of data to inform the algorithm what may constitute a high vs. low risk patient. In this scenario, there is synergy between a doctor’s inherent knowledge and experience coupled with the capabilities of AI to sift through massive amounts of data.
Decision support: Doctors make decisions that are supported by diagnostic analytics. Data-driven insights are inculcated into human knowledge and expertise.
Precision medicine, accurate diagnostics, genomic sequencing, cost reduction and population health management are all areas within medicine that AI Is being utilized to improve patient health. In the realm of care coordination, big data analysis is imperative to offering patients the best care they need in a timely fashion. Whether it be algorithms based on creating decision augmentation or automation tools, hospitals and accountable care organizations across the ecosystem are leveraging AI to draw insights from the data their patients generate. Recall, each patient interaction with the health system generates a node of data that when analyzed appropriately creates a wealth of insights into a healthier tomorrow.
For example, the Mayo Clinic and certain value-based Medicare Advantage programs currently use healthcare data analytics and AI data mining technologies to identify patients with more than one chronic comorbidity that may benefit from early intervention. By analyzing billions of data points, these algorithms assist in the coordination of care that would prevent these high-risk patients from needing to visit the emergency department. Although just one facet of how AI can coordinate care most effectively, the power of detecting patients at risk of hospital and ER admissions can save the patient and the healthcare system significant economic burdens, as well as potentially prevent the cascading of downstream sequelae of their disease state.
Overall, the power of using AI to not only analyze health data, but extract genuine insights that may impact patient’s lives is currently underway in our ecosystem. The coordination of care begins with robust and clean datasets, and superior AI algorithms are the perfect tool to offer doctors the necessary insights within these oceans of oftentimes scattered data.
Where hospitals can benefit from AI for care coordination
At the heart of true artificial intelligence is a composition of data engineering, data science, and machine learning. The structure and quality of healthcare data is the engine behind a clinically significant AI-based care coordination system. Prior to the advent of AI-based solutions, hospital care coordination was limited in the amount of data a system could not only hold, but make sense of. Without increases in computing power and machine learning algorithms, data in the healthcare system would continue to exist as fragmented bits of information that could not provide physicians, nurses, or health systems with clinically relevant knowledge. Through advances in interoperability, information exchanges, and data processing and mining, larger and more complex datasets are now utilized to guide decision-making. Care coordination AI is a prime example of how inculcating AI within the hospital can lead to improved outcomes, better care, and improved clinical decision-making.
Clinical Decision-Making
Decision support tools at the physician level have been utilized throughout the healthcare system for over a decade. Whether it be best practice alerts or embedded notification systems on the given medical records system, algorithmic decision making processes focused on taking the next best step in clinical management has proven to improve patient outcomes and decrease cost. For instance, a primary care provider may be managing a patient with suspected breast cancer. Prior to referring the patient to a medical oncologist, the PCP may be supported by a clinical alert system to have an ultrasound completed or specific lab tests run. Additionally, a patient presenting with acute lower back pain may desire to receive MR imaging, but clinical guidelines may reinforce that the PCP provide another less expensive and burdensome option prior to advancing with imaging. How may AI-based solutions augment this process?
Data mining: processing large sets of data enables predictive models to determine what the best course of action may be for a given patient profile. Meaning, a patient with congestive heart failure and three emergency department visits for pulmonary edema will be treated differently than a patient with pneumonia, although both are presenting with a chief complaint of shortness of breath. AI-integrated solutions can learn from hundreds of thousands of patient encounters, trace what outcomes were better than others, and provide insights accordingly.
Volume and learning: a physician learns from their patient encounters over the 3-4 decades they practice medicine. Each encounter informs them of what interventions work best, and how patients fare in the long term. Care management and care coordination AI tools can learn from 10-100x patient encounters, and provide answers to questions that a single practicing doctor would not be able to arrive at based on their clinical experience alone.
An initial impediment to turning big data in healthcare to actual informatics was the processing time required to translate data into actionable information. Even on a small scale, sifting through patient data can be daunting for a nurse or physician interested in extracting clinically pertinent information. Evidence-based medicine consistently demonstrates superior outcomes when compared to health systems or physicians that may not be leveraging some sort of data to guide their clinical practice. Consequently, the general consensus from hospital administrators and accountable care organizations is that medical doctors should be staying up to date with the most relevant clinical recommendations in their given specialty, and treat patients with data that is up to date. The attractive part of care coordination AIis its ability to shorten the time needed to gather, analyze and interpret patient data. Whether it be in the emergency department or the reading room, how time is spent within the walls of a hospital is of paramount importance to efficiency, outcomes, cost and throughput. Care coordination AI alleviates many of the pain points associated with actually integrating data in the decision-making process. By expediting the process by which data is processed and viewed, care coordination AI offers a method by which doctors can expediently learn about their patients and arrive at the best model of care for them.
Resource Utilization
The methodology by which different specialties communicate with each other in a hospital setting is inconsistent. Whether it be inputting consults into an outdated EMR, or quickly texting the cardiologist for his clinical opinion, forming a connection between specialists in a concise and timely manner is paramount to successful interdisciplinary medical care. AI-powered hospital care coordination creates an arena by which the radiologist and interventional cardiologist can both access the same information, while simultaneously sharing their thoughts and medical judgements with one another.
Furthermore, an effective care coordination model increases resource utilization within the hospital by engineering a structure that enables call-activation to be seamless. Through interoperability and information sharing, vascular surgeons and interventional radiologists can be informed of pertinent cases immediately entering their hospital. The intervening team can be at a patient’s bedside faster. This further improves the management of patients, as the appropriate type of doctor will handle the needs of the patient in a precise time frame. Hospital care coordination ensures that the right type of doctor presides over the care of each patient.
Economic Value of Care Coordination
The standards for AI solutions to become a vital part of conventional medicine are quite high. A novel technology must not only solve a clinically relevant problem, but it must also prove to be affordable to the patient population it aims to serve, as well as the healthcare system as a whole. Introducing care coordination systems that increase the overall cost of healthcare administration would severely limit its potential in being adopted by the marketplace. What makes hospital care coordination so attractive to both physicians and hospital administrators is its proven ability to improve patient outcomes while driving down the cost of care. It is self-evident that by driving the right physicians towards the care of an acutely ill patient that the patient’s clinical outcome will improve.
Though potentially overlooked, in tandem with clinical advantages are the economic benefits of AI in the healthcare system.Hospital care coordination that is powered by AI offers economic value in that more patients downstream are ensured to receive the most optimal care for their presenting symptoms. Moreover, by avoiding suboptimal clinical care pathways, hospitals may avoid preventable readmissions, which costs the health system and the Centers for Medicare and Medicaid Services a significant amount of money.
The AI-Driven Pulmonary Embolism Response Team
It may be odd to believe that chances, as a patient, of being operated on for a pulmonary embolism depends not on clinical acumen but on communication. Pulmonary embolism is the third leading cause of cardiovascular death in the United States, with up to 100,000 deaths annually. The majority of these deaths are caused by intermediate to high-risk PEs, with mortality rates ranging from 5% to 65% depending on severity. The treatment of patients with massive and submassive PE remains controversial. Different specialists bring different experience, technical expertise, and therapeutic recommendations towards the treatment and management of patients with PE. Therefore, to provide the highest quality of care for patients with PE, a team approach is necessary.
The Pulmonary Embolism Response Team (PERT) was established to individualize the management of PE and guide the use of effective treatment modalities. New approaches and therapeutic tools have shown promise in the treatment of high-risk patients with PE, including catheter-directed thrombolysis, improved surgical procedures, and extracorporeal membrane oxygenation (ECMO). However, determining the appropriate clinical care pathway for each individualized patient in real-time involves coordinated consultation and decision-making from a variety of specialties, including interventional cardiology, pulmonology, diagnostic and interventional radiology, emergency medicine and critical care. The less fragmented clinical decision-making is within each siloed specialty, the greater the chances that the appropriate and most effective therapy will be delivered to a PE patient. Pulmonary embolism care coordination ensures that all patients deserving of advanced medical interventions are offered the most appropriate life-saving treatments.
Moreover, PE care coordination provides a succinct and HIPAA-compliant platform for the immediate exchange of pertinent information vital for clinical decision making in the treatment of a pulmonary embolism. AI for PE care coordination affords physicians from all disciplines the opportunity to review CT angiography with the diagnostic radiologist, while obtaining troponin levels, blood pressure, and other important data points such as the RV/LV ratio. Currently, a significant portion of PE patients are assessed in the ED without input from interventional radiology. These patients may undergo suboptimal care due to a lack of consultation between the emergency department and other service providers, such as vascular surgery, diagnostic and interventional radiology, and critical care physicians. PE care coordination affords patients with the best health outcomes by guaranteeing that a heterogeneous set of subspecialists will review their cases, including but not limited to their CT imaging, RV/LA ratio, pertinent laboratory values, and overall clinical picture. What is most important about PE care coordination AI is its overall ability to improve the time to diagnosis and time to treatment of PE. Activating the interventional radiology or vascular team occurs expeditiously with AI-care coordination, and the patient is offered the most appropriate treatment in a more timely fashion.
Stroke Care Coordination AI
When dealing with neurons, every second counts. In the landmark SWIFT PRIME trial, Goyal et al. describe that if patients with ischemic stroke are reperfused within the first 150 minutes of symptom onset, the probability of functional independence is 91%. Importantly, this decreases by 10% in the first hour beyond the 150 minutes, and decreases another 20% with each subsequent hour of delay in intervention. Clearly, although team activation time is important throughout the hospital system, it is perhaps the most vital aspect of a stroke team. Currently, communication and information exchange between specialists within a hospital offer room for improvement. Moreover, communicative capabilities between primary stroke centers and tertiary care facilities – typically working in a hub and spoke model – may face obstacles when attempting to expediently transfer patients to the most appropriate site of care. Therefore, leveraging a care coordination stroke system that facilitates the immediate generation of vital patient data, including their brain imaging findings, may improve door to needle time for stroke patients.
Through eliminating unnecessary steps in communication and workflow diagnostics, stroke care coordination expedites tasks that can ensure a stroke patient undergoes endovascular therapy treatment in a timely manner. Utilizing stroke care coordination AIplatforms offers physicians the ability to be immediately available for intracranial procedures while offering a hub for interdisciplinary collaboration between neurosurgeons, ED physicians and interventional neuroradiologists.
Aidoc’s AI for care coordination
The vision of Aidoc is simple: improve patient outcomes with the power of AI. The Aidoc AI care coordination platform is a revolutionary innovation that focuses on four domains: 1) Accurate flagging of acute pathologies on diagnostic imaging 2) Inculcating relevant patient data in a quick and seamless manner 3) immediately informing the appropriate specialists needed to manage this specific pathology and 4) providing a platform through which doctors can discuss the patient and share information freely. The Aidoc Care Coordination system is an all encompassing technology that disrupts any siloed domains within the hospitals, such that communication between physicians and management of acutely ill patients are drastically improved. Whether it be involving interventional neuroradiology and neurosurgery for the treatment of intracranial hemorrhage, or providing a platform for interventional radiologists to discuss PE cases with their colleagues in the emergency department, the Aidoc AI care coordination system is the perfect solution for any hospital system aiming at improving quality of care while reducing overall cost.
