Using TeleICU powered by AI to combat the Coronavirus epidemic
Predictive and proactive measures — Artificial Intelligence can guide timely implementation measures allowing a reduction of both disease severity and workload
Remote patient monitoring — reduces the risk of clinicians’ exposure to infected patients providing protection for medical professionals
Minute by minute risk stratification — real-time acuity measurement can be used to determine timely interventions, improved prognosis for critically ill patients
Scalable infrastructure — allows healthcare providers to cope with a massive increase in ICU admissions
As coronavirus spreads, our resources are spread thinner
As COVID-19 rapidly spreads worldwide, healthcare resources are spreading thinner and critical care units face mounting pressure. We are facing a rapidly growing threat that demands a rapidly innovative solution given what we know about the COVID-19 virus. It has an efficient transmission rate (basic reproduction number, or R0 rate, of 2.0–2.2), meaning that for every person infected they have the ability to spread the disease to two more people. Both healthy and compromised patients with underlying comorbid medical conditions face a risk for mortality with a WHO estimated case fatality rate of approximately 3.4 % as of March 3rd 2020 (1,2,3,4,5). In addition, the propensity for the nosocomial spread of the COVID-19 virus mandates aggressive precautionary measures to control and prevent nosocomial transmission of the virus, given the abundance of co-morbid conditions present in the hospitalized population (3,6).
How will hospitals strategically leverage the limited resources at their disposal, and how can they integrate Telehealth and TeleICU services into their care mix to fight this epidemic?
Time for telehealth to step up to the spotlight
Last week the Center for Disease Control and Prevention (CDC) officials stated that the novel coronavirus (COVID 19) will spread in the US, meaning that hospitals, communities and individuals should ramp up their level of preparedness. In view of the expected increased demand for ICU beds, one action health organizations could take would be adding telehealth capabilities, an option that is indeed becoming a reality.
Statnews recently interviewed Eric Perakslis, Ph.D., a Rubenstein Fellow at Duke University focusing on data science. Dr. Perakslis, who had previously led the technology efforts for multiple Ebola response programs, stressed the benefits of using telehealth to protect medical professionals in triaging patients remotely and reducing the risk of medical professional infection. “Telehealth can be a force multiplier that helps protect health workers and extends their reach, and should absolutely be seized upon,” he said. Additionally, even at the point where a patient arrives at an emergency department, virtual platform initial assessment could help keep patients in isolation and minimize both clinical team exposure and viral spread within the hospital environment.(7).
Can artificial intelligence-based telehealth tools spur rapid innovation?
Telehealth and TeleICU options are useful, but as they exist today, they are not yet innovative enough to provide an entire solution to the coronavirus challenge (2) We need novel tools that allow for a predictive value as early as 6–14 hours in advance of complications or increased disease severity, in addition allowing for remote monitoring of patients via teleconsultation.
Using AI Tools with predictive and proactive measures to Innovate
Artificial Intelligence and predictive tools that can be used to provide minute-by-minute risk stratification and real-time acuity measurement are being tested in care settings today. They are being used to guide timely implementation measures that can reduce both disease severity and the load on the existing healthcare facilities.
Knowing who will progress from a cough to hospitalization, deteriorate to ICU admission, or develop a need for mechanical ventilation is as important as knowing who is expected to recover. AI-based telehealth tools enable this to be managed in a general ward or as an outpatient in a contained environment. This enables a patient to be kept away from already-crowded medical facilities and freeing up resources for higher acuity patients.
Here are some examples of how AI driven technologies can benefit in the management of COVID-19 patients:
We know that early isolation, diagnosis and management contribute to reduced mortality (2) Advanced AI tools could assess and predict respiratory decline in a coronavirus-infected patient up to six hours before they develop shortness of breath or bilateral patchy lung involvement with ground-glass opacity appearance on chest X-ray examination (2). The fact that CRP is raised in most patients, yet lymphocytopenia and leukopenia are more prominent in severe disease, can act as some sort of prognostic biomarker (2). In addition, recent evidence of proinflammatory markers could potentially be used as an early prognosis predictor, as the expression of several markers such as interleukin-2R and interleukin-6 was shown to be significantly higher in severely affected patients, in comparison to moderate- or low-severity patients (8).
With the rapidly transmissible coronavirus, it is impossible for the human mind to calculate and produce the insights that artificial intelligence platforms can, within such a short time; AI driven tools may help to accurately triage patients both in and out of the hospital settings, rapidly amass epidemiological data and feed into central global networks to rapidly gain a more real-time picture of the global infection rate, all while healthcare facilities fight a battle under limited resources and growing pressures.
Digital health tools originally designed for the flu are already being turned into triage techniques for coronavirus (6). However, the payment for such telehealth services is still hampered by state licensing regulations and the fact that insurers may not pay for it fully or in the same way. In the outpatient setting, AI driven cost-effective tools could be incorporated into home kit development that allow virtual monitoring of variables linked to prognostic biomarkers, in order to triage patients and channel the flow of higher risk individuals to medical care facilities.
Rapid uptake of AI needed for Predictive analytics in the inpatient setting
In the inpatient setting, hospitals are gearing up for massive scale ramp up efforts. An American Hospital Association-commissioned presentation warns that COVID-19 could create a disease burden 10 times that of influenza in the US with a projected 96 million cases, 4.8 million hospital admissions (1,7) million ICU admissions and 480,000 deaths. In the worst-case scenario hospitals and ICUs may run critically short of beds. The challenge is predicting who will deteriorate or improve to streamline patient flow. Here artificial intelligence big data algorithms are able to add significant value through their inherent capabilities to amass meta-data, incorporate evidence-based acuity scoring systems and provide rapid and real time assessment of individual patient variables. These AI systems thus allow clinicians the opportunity for intervening early and preventing deterioration, as well as increasing earlier discharge of low-risk patients from beds needed by more severely ill patients, and helping hospitals prepare for critical bed shortages.
Such technologies are already being considered and implemented by ICUs for critically ill patients in general. Without disrupting existing EMR systems, solutions such as those from CLEW sit as an additional information layer on top of the EMR system, pulling key patient monitoring parameters and providing real time minute-by-minute monitoring and real time views of patient health status within the last few minutes to hours. This multi-sourced information is subjected to an already built-in proprietary AI algorithm, developed in consultation with ICU specialists, and rooted in evidence based predictive scoring systems and best practice guidelines. The challenges posed by COVID-19 could expand the scope of such technologies beyond the ICU, and into the general wards, emergency departments and outpatient units of hospitals as well as into key primary care settings.
Are health care systems prepared for rapid innovation or ready to face a pandemic rapidly escaping our containment efforts?
In the short time since its emergence, COVID-19 is forcing us to re-examine the telehealth tools already at our disposal and by its very nature, acted as catalyst for remote monitoring and intervention. Innovative telehealth and TeleICU systems that incorporate AI can help healthcare services scale to address the vast need resulting from rapid disease spread and sustained transmission, while improving triage and early intervention, thus saving lives and reducing the impact of this pandemic.
Healthcare leaders need to ensure that their innovation strategies help them stay ahead of the developing challenges, and evolve at a rate that is at least as fast as the enemy they fight.
1. Gates B. Responding to Covid-19-A Once-in-a-Century Pandemic? N Engl J Med 2020;1–3: DOI:10.1056/NEJMp2003762. https://www.nejm.org/doi/full/10.1056/NEJMp2003762
2. Guan W, Ni Z, Hu Y, Liang W, Ou C, He J, Lui L, et al for the China Medical Treatment Expert Group for Covid-19. Clinical Characteristics of Coronavirus Disease 2019 in China. N Engl J Med 1–8: DOI:10.1056/NEHMoa2002032.
3. Habibzadeh P, Stoneman EK. The novel coronavirus: A bird’s eye view. Int J Occup Environ Med 2020; 11:65–71. Doi10.15171/ijoem.2020.1921.
4. Fauci AS, Lane HC Redfield RR. Covid-19 — Navigating the Uncharted. N Eng J Med 1–2. DOI:10.1056/NEJMe2002387.
5. Coronavirus ( COVID 19) Mortality Rate. Available from: https://www.worldometers.info/coronavirus/coronavirus-death-rate/ Accessed 8th March 2020
6. Munster VJ, Koopmans M, van Doremalen N, van Riel D, de Wit E. A Novel Coronavirus Emerging in China- Key Questions for Impact Assessment. N Engl J Med 2020; 382(8):692–694.
7. Ducharme J. The Coronavirus Outbreak Could Finally Make Telemedicine Mainstream in the US.
8. Chen L, Lui HG, Lui W, Lui J, Lui K, Shang J, Deng Y, Wei S. Analysis of Clinical Features of 29 Patients With 2019 Novel Coronavirus Pneumonia. Zhonghua 202;43:E005 DOI: 10.3760/cma.j.issn.1001–0939.2020.0005