When the World Health Organization declared Covid-19 a global pandemic in March 2020, hospitals saw a sudden surge of Covid patients in their emergency rooms. The overwhelming patient demand for scarce hospital resources grew at a high level they had never seen before.
Two years later, as the world wrestles with the next pandemic wave, governments are still very concerned about hospitals’ scarcity of ventilators and ICU beds. More importantly, governments are worried about the alarming shortage of a hospital’s most valuable resource: medical staff.
“The scarcest resource is not stuff; the scarcest resource is actually people,” maintains Julia Lynch, a University of Pennsylvania professor. Or, to paraphrase University of Michigan professor Deena Costa, it is easier to produce a ventilator than a skilled doctor.
The Covid pandemic has increased patient demand for hospital services while, at the same time, the hospital staff is getting sick. In addition, hospitals have had to cancel elective surgeries and other procedures to make room for Covid patients, further exacerbating the shortage of medical staff. The result is that hospitals are struggling to keep up with patient demand.
“In the Covid-19 pandemic, we are seeing a perfect storm of challenges to the health care workforce,” said WHO Director-General Tedros Adhanom Ghebreyesus.
Using AI To Help Doctors Triage
There are more Covid patients than doctors in a hospital to care for them. Moreover, many doctors are considering leaving their profession after the pandemic due to physical and mental burnout. Thus, hospitals are adopting AI at an accelerated rate to address the shortage of doctors and their anticipated exodus from the profession.
Covid-19 patients are typically triaged by medical staff to determine the severity of their illness and what course of treatment is needed. However, due to the shortage of medical staff in many hospitals, AI systems are being used to automate this process.
Many hospitals have employed AI to perform triage decisions during the pandemic. Its growing ability to make autonomous decisions faster, better, and cheaper than humans profoundly impact the medical field.
Medical AI is rapidly improving its performance in recognizing patterns and diagnosing diseases more accurately. As Dr. Eric Topol points out: “Unlike humans who get tired, have bad days, may get emotional, sleep-deprived, or distracted, [AI] machines are steady, can work 24/7 without vacations, and don’t complain.”
Troubling Triage Takes Toll On Doctors
With increasing patient care demands far outstripping available hospital resources, doctors are compelled to make difficult triage decisions, that is, how they can fairly allocate scarce medical resources to many patients during the pandemic.
When doctors give ventilators to patients, those decisions “translate into who is given a chance to live… and who will be left to die a complicated death.”
Triage decisions like these have contributed to their mental distress leading to doctor burnout. It was reported that Italian doctors were “weeping in the hospital hallways because of the choices they were going to have to make.”
A Shift in Medical Ethics Practice
A key factor causing mental distress is that doctors are forced to shift from practicing clinical ethics to public health ethics abruptly.
Before the pandemic, doctors employed clinical ethics, making them advocate for their patient’s well-being. Patients who needed and wanted to use a ventilator had access to one. Likewise, withdrawing a ventilator would depend on whether the doctor no longer recommended it for the patient or if the patient requested its withdrawal.
During the pandemic, doctors must prioritize public health ethics over clinical ethics. It is their ethical duty to “do the greatest good for the greatest number of patients” over and above the well-being of patients for whom they care.
This moral duty means grappling with triage guidelines to determine how to save the most lives. They can no longer actively advocate on their patients’ behalf.
This tough decision-making has affected their mental health, leading to burnout. To support our doctors during this pandemic, we must remember that they are human, too. As such, they are susceptible to the same emotional turmoil we experience.
5 Principles To Ensure AI Triage Is Ethical
For all its promises, AI brings along some perils. How can hospitals ensure that they maximize AI’s benefits while minimizing its risks?
One way is for hospitals to use the global ethical AI framework comprised of five core principles. The framework includes the four classic bioethics principles of beneficence, nonmaleficence, autonomy, and justice, as well as a new principle of explicability, given the black box black-box nature of AIAI’s black box black-box nature. The European Commission and the OECD have used this framework in drafting their AI ethics guidelines.
How does a hospital apply these five principles to ensure that AI triage is operated ethically?
1. Beneficence
A hospital can uphold beneficence by responsibly stewarding its scarce resources. It achieves this by fairly applying the triage guidelines to achieve the greatest good for the most significant number of patients.
2. Nonmaleficence
The hospital can also promote nonmaleficence by preventing data privacy breaches of patient data. It reduces this risk by ensuring the security and confidentiality of patient information it collects, uses, and stores.
3. Autonomy
The hospital can also respect patient autonomy by not requiring patients to undergo medical interventions against their wishes. This is achieved by ensuring that an intervention does not contradict a patient’s values and beliefs.
4. Justice
The hospital can also advocate for justice by eliminating bias and discrimination in its datasets. Biased datasets can include race, gender, level of education, or disability.
For example, Tennessee’s original Covid triage guidelines discriminated against people with disabilities. Its guidelines excluded persons who did not have “long-term survival prospects” from accessing medical resources. The hospital eventually removed this limited exclusion from its updated triage guidelines.
5. Explicability
The hospital can ensure explicability by making its AI triage transparent. It achieves this by clearly communicating the criteria used in triage decisions to maintain patient trust in the health care system. The hospital can also ensure accountability by taking full responsibility for all triage outcomes.
By applying these five principles, hospitals can help ensure that AI triage is ethical.
Some Concerns Using AI in Health Care
However, some ethicists have raised concerns that using AI in health care may violate some of these principles. For example, they argue that AI’s black-box nature makes it impossible to explain how it arrives at its decisions. As a result, AI may violate the principle of explicability.
They also argue that AI may violate the principle of beneficence because it can lead to decisions that are not in the patient’s best interest. For example, imagine an AI system deciding which patients should be given life-saving medical treatment. The AI system may consider a patient’s age, health, and the likelihood of survival when making its decision. However, the AI system may not consider a patient’s wishes or values when deciding. As a result, the AI system may make decisions that are not in the patient’s best interest.
Despite these concerns, some ethicists argue that the use of AI in health care can be ethical if it is done in a way that respects the five principles of beneficence, nonmaleficence, autonomy, justice, and explicability.
The American Medical Association considers AI “not as artificial intelligence but as ‘augmented intelligence’ that enhances rather than replaces physicians’ expertise.”
“At the end of the day,” as Stanford doctor Ron Li confirms, “it will still be the human experts who will make the call regarding whether or not the patient needs to go to the ICU or get intubated — except that this will now be augmented by a system that is smarter, more automated, more efficient.”
Other Ways AI is Helping During Covid
1. Automated Diagnosis
Doctors at Wuhan Central Hospital in China were the first to use an AI system for Covid-19 diagnosis. The AI system could read CT scans 30 times faster than a human radiologist and with high accuracy. The system correctly diagnosed Covid-19 95% of the time, while humans had an accuracy rate of only 85%.
In high-risks, high-stakes care scenarios, AI is a “shared mental model” for doctors and nurses to agree on a patient’s diagnosis. This has helped reduce the mental distress of doctors determining who lives and dies.
For example, AI is now used as a support tool to help predict “which patients will have a cardiac arrest, which will need transfer to the intensive care unit (ICU), and which will die — all within the next eight hours.”
AI is also used to review radiology images that reveal lung abnormalities associated with severe Covid cases with promising 95% accuracy. It can also triage patients for possible Covid infection risks in 10 minutes instead of 12 hours.
AI systems are also used to help doctors select the best treatment for Covid-19 patients. In one study, an AI system could correctly predict the most effective treatment for Covid-19 patients 80% of the time.
Using AI for Covid-19 diagnosis is essential because it can help hospitals triage patients more efficiently. With the pandemic putting immense strain on the healthcare system, AI can help doctors and nurses by taking on some of the tasks that would typically fall to them.
2. Nurse Robots & Virtual Nurses
Many hospitals use nurse robots to disinfect rooms and deliver patient meals. They are also using AI-enabled virtual nurses powered by AI to help answer questions from patients and their families about Covid-19.
There are many Covid-19 questions that patients and their families have, such as:
- How do I know if I have Covid-19?- What are the symptoms of Covid-19?
- How can I prevent getting Covid-19?
- What should I do if I think I have Covid-19?
- How can I care for someone with Covid-19 at home?
The virtual nurse powered by AI can provide answers to all of these questions. Additionally, the AI can keep track of all the questions that have been asked and use this data to improve the quality of the answers.
This is important because hospitals can provide better care to Covid-19 patients and their families. It also helps to free up the time of human nurses so that they can care for other patients. AI for virtual nursing is an example of how AI can be used ethically to improve healthcare quality.
3. Contact Tracing
AI is being used to help with contact tracing of Covid-19 patients. The technology rapidly identifies close contacts of infected patients and alerts them so they can self-quarantine and get tested.
This is important because it can help to prevent the spread of the virus, and it can also help to identify asymptomatic carriers. Asymptomatic carriers have the virus but do not show any symptoms. They can unknowingly spread the virus to others.
AI-powered contact tracing can help to reduce the transmission of Covid-19 by identifying asymptomatic carriers and alerting their close contacts so they can take precautions.
4. Fever Screening
AI is being used to screen people for fever at hospitals, airports, and other public places. Thermal cameras equipped with AI software can scan large crowds of people and quickly identify those with a high fever who may have Covid-19. This is important because it helps triage patients and ensures that those most likely to be sick are seen by a doctor first.
However, there are ethical concerns about using AI for fever screening. Some believe it is inaccurate enough and could unnecessarily lead to quarantining healthy people.
There are also privacy concerns, as thermal cameras can capture images of people’s faces. Privacy experts have raised concerns that this could lead to facial recognition being used to track people’s movements. AI can be a valuable tool in the fight against Covid, but only if it is used responsibly.
5. Drug Discovery
AI is being used to speed up the process of discovering new drugs to treat Covid-19. AI systems can rapidly analyze vast amounts of data to train algorithms to predict potential drug targets and accelerate drug development. This is important because it can help get new treatments to patients faster.
There are many ethical considerations when using AI for drug discovery, such as ensuring that the data used to train the algorithms is representative of the population that will be treated with the new drugs. Governments must be aware of these considerations and ensure they are ethically sound before using AI for drug discovery.
AI’s Greatest Opportunity in Health Care
When used in these ways, AI has the potential to improve patient care by providing physicians with more information and insights that they can use to make better-informed decisions. However, it is essential to note that AI is not a panacea. It cannot replace the expertise of experienced physicians. Instead, AI should be used as a tool to augment the decision-making of physicians.
Dr. Topol summed up AI’s ultimate benefit to the medical profession:
“The greatest opportunity offered by AI is not reducing errors or workloads, or even curing cancer: It is the opportunity to restore the precious and time-honored connection and trust - the human touch - between patients and doctors.”
We do Algorithmic Impact Assessments. Contact us!