August 1, 2020
The COVID-19 pandemic is shaking our societies to the core. In no time, the virus has painfully brought to light insufficiencies and discrepancies in healthcare accessibility and quality around the world. As we move further into the crisis, other more insidious health implications become apparent, starting with the psychological impact of anxiety and social distancing on mental health. In the face of these challenges, innovative strategies are being put into place, among which data and artificial intelligence have a leading role to play.
Joining international efforts to fight the current crisis, “the community of AXA researchers has pulled together, even closer than before, to inform and shed light on the pressing issues raised by the current crisis,” reports Marie Bogataj, Head of the AXA Research Fund. Among other initiatives, the fund has brought together nine of its most prominent chairholders to form a COVID-19 Task Force.
In this third and last of a three-part Expert series, three top-tier AXA-supported researchers provide their multidisciplinary insights into health systems response, the role of data and AI and on mental health effects.
Weaknesses and inequities in our social systems have existed for years. With the current crisis, they have become exacerbated and thus more visible. I will give you a few examples, first regarding global guidelines on how to prevent the spread. One recommendation is that people should stay at home. Not everyone can. In Mexico, for instance, 49% of the population does not have enough income for basic food security. For them, to stay at home, is to stay hungry. Many people say the COVID-19 crisis will be a hunger pandemic. Another example concerns education and the recommendation to follow courses online. Here in Mexico, only 56% of homes have Internet and the more disadvantaged students are likely to fall even further. Another recommendation is to wash your hands as often as possible. In Mexico, 54% of homes don’t have a daily supply of water. These guidelines have been designed for developed countries. They have not been adapted to other countries, or even to people within the developed countries who cannot follow them. They are exacerbating inequalities, and making more difficult to control the pandemic.
In terms of health system, not all countries have universal health coverage. In Mexico, 58% of people don’t have formal coverage, public or private. In New York, the mortality is much higher in Latino and African American communities. This is due, among other things, to the differences in access to healthcare. This pandemic has also made very evident that parts of the health system are much more neglected than others. Nursing homes, for instance, were not prioritized even though we knew that the people in them were more at risk.
I am a very strong advocate of this approach. "One Health" is officially defined as a collaborative, multisectoral, and transdisciplinary approach—working at the local, regional, national, and global levels—with the goal of achieving optimal health outcomes recognizing the interconnection between people, animals, plants, and their shared environment. The spread of the coronavirus, which we know jumped from wild animals to humans, was made possible by environmental changes and interconnections between human societies and countries. The idea behind the "One Health" approach is that you cannot understand the crisis by just looking at human health. You have to look at all the other components and think of them collaboratively. Countries must acknowledge these facts. Economic and scientific collaboration between countries, not isolation, is the only way forward.
We are currently urging government officials to consider the principle of One Quality for all types of health services, not just in terms of clinical care, but also in terms of public health services. Clinical care is just one part of the problem, like trees preventing us from seeing the forest. The forest is the One Health approach. Public health services are part of it all, and we need to apply the same quality requirements and quality improvement actions to them as well: one health, one quality.
To prepare for the next pandemic, I think there are five important things to think about. The first one is the most obvious: prevention. We need to commit to "One Health" initiatives and consider them as an investment for the future, and not so much as an expenditure. If we had invested in One Health prevention, the economic losses would not be as bad or may have even been inexistent, because the pandemic would not had happened in the first place. The second thing we need to think about is the protection of our health systems, and particularly human resources. In Spain, for example, 20% of front-line health professionals have been infected so far, probably due to inadequate or inexistent protection. The third is infrastructure. We have to anticipate what we may need if this type of situation happens again and think about what our responsibilities are as a health system. The shortage of protective garment worldwide is a striking example. Now that we are waiting for a vaccine, we need to make sure that it will be safely distributed everywhere and for everybody. Again, this is obvious, but are nations thinking about being prepared for it? The fourth is ethics. Think of Italy. When they became overwhelmed with patients, they did not know what and who to prioritize. We need to agree, in advance, on an ethical statement establishing rules for prioritization. The case of health personnel makes this obvious; their safety should have been a priority. Fifth, and finally, the important issue of fighting inequalities: social inequalities such as access to clean water and education, and also health inequalities and universal access to quality services. Fighting social and health inequalities should become a real priority.
I think that the first thing that is important for us to discuss is why this period is so stressful. The thing we know about stressors is that they are especially dangerous to us if they are uncertain. Right now, we don’t know if we will get sick, if we will have a minor set of symptoms, or if we will be in the hospital struggling for our lives. We don’t know if we will lose our jobs, or if our basic needs will be met.
The current situation is also what we call an uncontrollable stressor. Whether or not we are stuck at home is not up to us. We cannot do anything personally to remove that stressor. That is why you see people behaving in these odd ways like buying up all the toilet paper in stores. They are trying to get control over something. In the short term, stress is not that harmful for us. Our bodies are built for short-term flight-or-fight situations. But when the stress keeps going for a long-extended period, like what we are going through right now, what happens is our bodies start dysregulating. They fail to go into homeostasis and that’s when you start seeing both psychological, behavioral and physiological damage happening.
We start engaging in unhealthy coping behaviors. We overeat, we smoke, we drink heavily…. And then, to make matters worse, we cannot engage in our good coping habits. For example, we cannot go to our gym and we cannot see our friends. We know the harms of smoking and drinking, but few people know the harms, for instance, of not taking a vacation, which literally can kill you. This partnered with our bad health habits, is what is leading us to a trajectory which absolutely can lead to clinical depression, and a lot of us are worried about suicidal behaviors going up.
This pandemic is eroding our social relationships, not just outside the homes, but inside as well. Because we are stuck together all the time, our homes can become a source of stress as well. We know that, in some cases, these situations can result in violence.
I wouldn’t go as far as to say that the cure is worse than the disease, but we do have to think about the mental health consequences of the lockdowns. In order to adapt to the anxiety this situation is creating – that constant feeling like a lion is about to attack us – the best intervention is to try and release the energy our bodies are saving up, to release the adrenaline flowing through our bodies. We can do some exercise: find some high-intensity workout online or go for a run if we are allowed to.
Another issue from a public mental health perspective is the inability of people to go for walks or let children outside. Nature and exercise are huge well-being predictors. Just being around trees can reduce your blood pressure. If you cannot exercise, because you are physically unable to, even simple breathing exercises can help. Paced breathing can activate the relaxation component of our nervous system and physically lower your heart rate and blood pressure. From the social relationship perspective, capitalizing on the relationships that we have can help, like reaching out to people online. I have been telling people to try and concentrate on the positive things they have now and the ones they can plan for. I would also recommend that people just look up positive psychology interventions online, there are so many great things out there.
From the public health perspective, we could start thinking of slightly larger bubbles of isolation. We could let people quarantine together, in two-or three- family groups for example, in which people can share childcare duties across families.
Our minds and bodies are not dual systems, which means that anything that we feel or think has the potential to alter our physiology. We have decades of research showing how our mood, our stress and our relationships affect our stress hormones, how our immune system works, how our cardiovascular works, whether or not we recover from an illness. We also know that protective factors, like social relationships and positive emotions, can undo those stress processes and protect us so that we don’t get sick. We really understand those mechanisms quite clearly.
There are so many important psychosocial and behavioral things that we need to look at to predict COVID-19 outcomes right now. What this means for research is that, the huge variability we are seeing in how sick people are getting, beyond preexisting conditions or age, could be partly explained by psychosocial predictors. We need to understand to what extent the behaviors we are having at home right now, how we are acting in our relationships, our thoughts, our stress, are influencing not just whether we catch the virus or not, but how sick we will get. If we can figure this out, we can tailor psychosocial interventions over the coming year.
Prof. Thomas Lukasiewicz - Yes, of course. AI can help to detect patterns that belong to certain mental health diseases, like depression. We often express our feelings in our face, but also our voice. These patterns of diseases can be detected. Wearable sensors can also help detect relevant variables. The same way, online activity can be used as a detection opportunity. Chatbots could also play a role, they could help provide support to people who have nobody to talk to, and, of course, help detect pointers of depression in the speech.
Prof. Sarah Pressman - I think what is going to be particularly useful is social media. Obviously, what we say online, the words that we use, are different than what we feel internally, but there has been so much exciting work showing that you can pick up distress by looking at the words that people use online, and even connect that to health. I know that places like Facebook have developed algorithms to scan posts for suicide risk using machine learning of unnatural language processing. Hopefully, we will see people partnering with public health organizations at some point to potentially intervene when we do start seeing these behaviors. Another thing that has been key is the deployment of telehealth approaches, both MDs and clinicians. Many people were able, almost right away, to move their practices online. It is going to be so important for us monitoring mental and physical health. Hopefully, it will become more and more available for everyone.
Prof. Thomas Lukasiewicz - There are several examples that illustrate how we can use AI and machine learning, more specifically deep learning and pattern recognition, to provide alternative ways for testing and diagnosing Covid-19. We have all heard of the Chinese clinicians using X-ray imaging and CT imaging to diagnose the virus. These images are currently being used in order to build AI systems capable of detecting it. Since the virus is affecting breathing, and contaminated people are often coughing, another idea is to use AI to detect the virus by identifying patterns in the voice. Wearable sensors could also be used, that record our life and mobility data. People with the virus might exhibit higher temperatures, they might not sleep well during the night or exhibit other deviations from their normal behavior. If we take a broader look at the problem, away from physical symptoms, AI can also be used to detect new outbreaks using people’s activity online. People may mention relevant information on social media, or they may research symptoms on search engines. Machine learning is also used for treatment. As of yet, there are not many possibilities for treatment, but AI can help decide if ventilation is necessary, when it is necessary, under which conditions. It may help predict and optimize outcomes to increase the chance of survival of patients. AI is also being used to design drugs and vaccines to try out different molecules and their effects. Later, for when we have tackled the pandemic, AI could also help us to monitor the long-term impact of the virus on the cardio-vascular system for instance.
Prof. Thomas Lukasiewicz - The big questions are, first, which data to store and, second, whether to store it in a centralized or decentralized way. On the issue of what information to store, what we really need if someone gets infected and tested positive, is to inform all the previous contacts that they are at risk of infection. We need those contacts, but we don’t really need their locations. There are solutions that offer complete tracking of these people, including their locations. This is more information than we need at that point. Now, on the topic of where to store this information. If we stored all the information in a centralized way, we could use that information for other purposes, like exploring the most likely transmission circumstances, but that would also open a way for misuse. That is why people, and scientists, would prefer to decentralize the information.