PART I. Ai and COVID-19- How Ai Can Provide Decision Intelligence For Pandemic Response

image credit cdc.gov

Amidst the COVID-19 pandemic, we are now receiving data and reliable research from extensive scientific efforts in Mainland China and Europe.

Given my business exposure to APAC, I personally had my family establish Asian mitigation protocols in the face of this virus, prior to any western measures that were instituted. Singapore and Hong Kong provided hope and many lessons to the world in the presence of COVID-19, after they learned their lessons in the SARS (SARS-CoV-1) epidemics of 2002–2003.

As a Data Scientist and Economic Mathematician, the overall impact of this public health crisis and the increased interoperable network effects not only to our daily lives but to our economies, and how we will make the decisions we will make in the near future. The new paradigm as individuals and a community are not only enormous, but present us with a new fiscal, policy, and economic standards that as a collective we haven’t faced.

That is why I felt the ethical responsibility as a scientist working in 4IR technologies and Ai to write this two-part article. It is essential now more than ever that technologies like Machine Learning and Artificial Intelligence are appropriately positioned to the public so we can understand as a collective that these technological advances can help us. Ai is not just something tech companies do, or Data Scientists excel at. Ai, and all the technologies associated with this family of methods like Machine and Deep Learning, are technologies that impact us all. This is why if you are a corporate leader or in the public sector, you must realize the impact of these technologies and how much they are changing the way we do everything. In the corporate world, Ai will give an advantage only to those first movers. In the face of this pandemic, we must understand how these advances ethically implemented, can enhance our human cognition, and assist us in a human and machine loop. And in all of the chaos and negative externalities this COVID-19 pandemic presents, it also offers the opportunity for us as a society to see the power of a Human Centric Ai.

Ai applications are currently being developed and have already been tested for COVID-19 in Asia. We are already seeing data being modeled by Machine Learning and Ai algorithms collaborating to a solution on a variety of fronts, and most importantly, aiding in understanding this novel pathogen. Understanding is the crucial point. Knowledge of the COVID-19 behavior of contagion, the epidemiological front, understanding of the human behavior in the presence of this pandemic, and understanding the impact of further informational asymmetries in governments around the world, are crucial to developing proper policies. These policies can be implemented for community-based strategies to reduce contagion spread, safeguard our health-care infrastructures, and mitigate downside to our economies.

Ai and COVID-19 Diagnostics and Treatment

As an Ai community, we are seeing functioning Ai methods in diagnostics and treatment. The truth is that this is a clear area of clinical attention given that the quantitative evaluation method of lesions involves multiple factors such as the cumulative lung volume range and density of the lesions, and there is currently no uniform standard. Additionally, even though pathogenic laboratory testing is the diagnostic gold standard, it is also a time-consuming process with a significant proportion of false negatives. Without CT-Ai COVID-19 real-time recognition models, this is a manual method requiring several hours of manual evaluation, with unfortunately low efficiency in human recognition of the lesions and COVID-19 pulmonary structures.

Professor Yuxin Shi, the Deputy Director of the Shanghai Public Health Clinical Center (SPHCC), guided the development of such Ai system with a private-venture structure, and as he states that “CT imaging is one of the important diagnostic and therapeutic basis for COVID-19, and it can quickly realize the diagnosis of viral pneumonia”. Under his guidance, the SPHCC is the first Ai imaging product in the industry to intelligently evaluate COVID-19, as reported by the HIMSS Asia Mobi Health News this past February.

There has been a clear and promising proof-of concept with Ai-Deep Learning leading the efforts to extract radiological features for timely and accurate COVID-19 diagnosis. There are several factors that push for a prompt productized Ai solution: since before the pandemic status, the efficiency of the nucleic acid testing has been seriously dependent on several rate-limiting factors, including but not limited to the availability and quantity of the testing kits in the given infected area. The fact that screening logistics of large numbers of suspected cases for appropriate social separation/quarantine and treatment is a pressing priority; also a matter of concern is the factor of the quality, stability and the reproducibility of detections kits; the impact of methodology, disease stages, specimen collection methods, nucleic acid extraction methods, and the amplification system are all determinant factors for the accuracy of test results. Additionally, conservative estimates of the detection rate of nucleic acid are low and tests need to be repeated several times before they can be confirmed.

There’s an opportunity in this COVID-19 pandemic to exact the power of Ai-diagnostic deployment in our favor as a human collective, as radiology imaging is also a major diagnostic tool for COVID-19. Radiologically speaking, the majority of COVID-19 cases show similar features on CT imaging inclusive of ground-glass opacities in the early stages and pulmonary consolidation in the late stage. Additionally, the images of various viral pneumonia are similar and do show overlapping with other infectious and inflammatory lung diseases, thus making it difficult for radiologists to distinguish COVID-19 from other viral pneumonic conditions.

In SPHCC the enhanced Ai-power is being given to clinicians with a system that uses Ai diagnostics and quantitative evaluation of CT images of COVID-19, while also grading the severity of various pneumonia diseases of local lesions, diffuse lesions, and whole lung involvement. This Ai-algorithmic solution is also accurately quantifying the cumulative pneumonia load of the disease through quantitative and omics analysis of key image features such as the morphology, range, and density of the lesion. It also aids in the clinical medical judgment of the condition while evaluating and giving prediction of prognosis, via a dynamic 4D contrast rendering of the whole lung lesions on CT. And as reported by the HIMSS Asia Mobi Health News, “the system can complete a quantitative analysis of lung lesions in 2–3 seconds”.

In the West, with the collaboration of a coalition of governments, health-care and public health institutions, and the private sector, we can develop a centralized repository and retained learning Ai engine to help in the fight of the COVID-19 pandemic, and overcome the obstacles and challenges this disease throws at us together, beyond borders closures, and as one humankind.

Google’s DeepMind is taking a step towards this, by applying its existing work to help the scientific community in the fight against COVID-19 with their research aid in the releasing of structure predictions of several understudied proteins associated with COVID-19, with DeepMind’s Deep Learning system AlphaFold.

Ai and COVID-19 Epidemiological Tracking and Forecasting

On the epidemiological tracking and forecasting front, Ai is showing a tremendous amount of promise and effectiveness as well. I think it is here, where people will likely have the most of the sci-fi feel of Ai in some of the applications. This is the case of a US-Hong Kong venture which was tapped to help Hong Kong doctors and researchers combat the COVID-19 outbreak via an Ai and remote monitoring platform. According to Professor David Chung Wah Siu, M.D. of the Department of Medicine at the University of Hong Kong, by being able to collect all the clinical labs, and physiological data, the goal of this project is to increase the epidemiological understanding of COVID-19 in order to improve the outcomes and treatments of the people of Hong Kong as well as the global community as more people become infected with COVID-19.

Another application, deployed in mid-February as reported by HIMSS Asia Mobi Health News, was in Mainland China where aside the extreme quarantine measures “the Chinese government released a public app to gauge potential coronavirus exposures — the tool acts as a way to collect data as well as to educate citizens on what to do if they have been in close contact with the virus — when to stay at home and get advice from health authorities”. These mobile integrated Ai tools, provide COVID-19 apps with scanning and augmented reality capabilities including geofencing, allowing the user to know where they enter COVID-19 risk zones. The tools take people’s temperature, among other features, with an overall result of helping communities and individuals make better decisions while at the same time not entirely halting the economy to a stop.

Mainland China is doing a lot more than just apps to help the population and governmental authorities in this new COVID-19 world. Based on reports from the Asia Times earlier this month, China has been leveraging cell phone big data, and Ai such as that “Chinese government algorithms can estimate the probability that a given neighborhood or even an individual has exposure to COVID-19 by matching the location of smartphones to known locations of infected individuals or groups”.

The reality is that all smartphones with enabled GPS give cellular and telecom providers a detailed and precise record of all the user’s itinerary and path of behavior. This has been a reality for years, and if you have ever heard me speak, I have touched upon this issue many times. With this information and elementary modeling and Ai, one can predict with extreme accuracy the user’s behavior and predicted location a month from now, for example — these are old news. However, what is new news is that the Chinese authorities are using this information to manage limited medical resources more efficiently. For example, directing tests for the virus to high-risk persons of interest identified by the Ai algorithm, as reported by the Asia Times. The leverage of big data analysis is giving Chinese authorities the power to establish — with high precision — the chain of transmission of the virus in any communicable epidemiological path, which remains a mystery in most other countries. But in China, without privacy laws that prevent collecting this data from users — like it is the case of the data privacy landscape we are used to and expect in the USA and Europe — the power of Ai is being used to analyze where the infected individuals had been during a six-week interval while identifying all the possible points of intersection and then requires tests to all the possible intermediate human carriers of COVID-19. Important to note is that the rapid speed of availability in the implementation and execution of Ai in China with telecom big data, is a clear indication that this 4IR field has been industrialized for quite some time. In the West, we need to find ways to enable the power of a Human Centric Ai to achieve these positive externalities in cases of the pandemic — as China has shown. Still, in a new way, by leveraging the voluntary contribution and availability of data without violating individual privacy safeguards, we have fiercely protected in our Western democracies.

On the other hand, Ai has also provided on this epidemiological front, ethically sourced methodologies and models with modified stacked auto-encoder methods for modeling the transmission dynamics of COVID-19. An approach I have investigated and that I find extremely promising is modified modeling with COVID-19 SEIR (Susceptible-Exposed-Infectious-Removed) and Ai prediction, trained in the 2003 SARS data and COVID-19. The modified original SEIR-equation takes into account a new dynamic of Susceptible and Exposed population in the parameters, given the data landscape of COVID-19 which for example considers the latent population — Exposed — to be infectious yet asymptomatic, which is consistent with the data from South Korea compared to the Italian data. Thanks to broad-based testing were able to effectively diagnose the 20–29 age group of the population as infected versus the latter, which only tested the very ill population. This latter approach uses Ai for prediction of the COVID-19 epidemic peaks. It should be pursued to provide better guidance to properly plan for public health measures indispensable in this pandemic times to achieve reduction and non-resurgence of COVID-19 infectious population.

Last but not least, another area of great Ai methodological application in our new paradigm of a COVID-19 world, is in the estimation of the risk of sustained community transmission of COVID-19 outside of Mainland China. Given its intersection with broader human impact, I will discuss this in the following section.

Ai and COVID-19 Broader Human Impact

Prescriptive and predictive analytics, the new areas beyond the traditional descriptive analytics — the bread and butter of traditional business intelligence — corporations are accustomed to, are possible with the advancement of overall Ai methodologies and models. Even at an analytics level, Ai can have a clear broader human impact on the COVID-19 pandemic — and pandemics in general — on economies. This impact can provide powerful assistance in information, while if not expertly and ethically implemented can also produce negative effects of dis- and misinformation or abuses thereof.

A great exemplary effort of the positive prescriptive analytics with Ai, was showcased in a free distribution article by the Washington Post on March 14th, using an open repository of data set in place by the Johns Hopkins University Center for Systems Science and Engineering. The simulations shown are a great example of prescriptive analysis that can be extremely helpful for policy decisions in terms of the extent of social distancing. Proper inference and analysis should prompt thoughtful policies promoting intermittent stringent social distancing over draconian quarantines, in terms of percentage of the population that can remain not-infected from COVID-19. These findings may seem counterintuitive — especially in the face of fear of COVID-19 infection, but these are the powerful Ai tools that can help propel and inform, not only our leaders but also our population, help ease individual overreactions and crisis-mode behavior, help better manage our supply chains, and help in not collapsing our much-needed infrastructures of health-care, commercial, and retail supplies chains, while ad unison mitigating the negative impact to our economy. I have personally forked and accessed this data and have started to analyze it as well.

With our US infection rates grossly under-reported due to the delayed governmental action and testing, it is questionable if those individuals currently testing negative for COVID-19 are to confirm the test result, given the scarcity of the main form diagnosis.

As mentioned in the previous section, there’s a concentrated effort in estimating the risk of sustained community transmission of COVID-19 outside of Mainland China utilizing Ai methods. The models are calibrated on international case importation while including modeling of travel restrictions within China and to and from international destinations. These models need to be recalibrated with as much data as possible predominantly from Europe and the USA, especially with more and more countries closing their borders. Epitomizing this is Chile — in South America — announcing on March 16th with a case count of 155 in a country with a 19 million+ population, a full border closure while alerting international aviation carriers that the initial term of the drastic policy will be in effect until April 30th. Our new COVID-19 world is prompting more and more countries to impose travel restrictions to specific origination localities with high levels of contagion, while we may also be facing more and more countries like Chile imposing full border sanitizing closures. Regardless of the need of model calibration in the ever-evolving data topography that COVID-19 brings forth, the results of these early year Ai models of risk estimation in sustained community transmission, suggest that at least a higher than 60% broad-based testing is necessary to avoid mass contagion, which unfortunately has been neither the protocol nor the reality in our Western world, so our projections even in a flattened curve scenario are not favorable. We need to urge our government’s broad-based testing for COVID-19.

The somber reality we are faced with as a human collective is that governments will not be able to minimize both the deaths and the economic impact of the viral spread of COVID-19. As I have informed in this article keeping mortality low involves governments putting in place measures that will inevitably spur economic turndown. As of the market close of March 16th, the Trump administration acknowledged that the mitigation measures of social distancing may be in place through July and August of 2020.

The importance of Data Science, Machine Learning, and Ai model-based predictions can be of invaluable help in these pandemic times for policymakers, even under the clear existence of the uncertainties of COVID-19’s mechanics. Additionally, behavioral economics, networks, and sophisticated models under uncertainty — integrated to Data Science, can also aid in the difficult decisions ahead for governmental actions as to understand how individuals respond to the advice and how best to prevent transmission. The aforementioned are also crucial in government communication strategies to keep the public informed into how to avoid infections best, as it is extra support to manage the economic downturn.

In Part II, I will explore further the broader human impact of COVID-19 and pandemics in general on economies, culture, and human behavior. Beyond a global economic turndown focus, Part II of this article will emphasize as to how the broader human impact of COVID-19 will affect human behavioral economics, spurring a new economic paradigm affecting current commerce infrastructure, impacting interoperability of networks, online traffic, and the expanded need for ethically developed and implemented Ai and 4IR technologies deployment.

About The Author:

Sol is a Mathematical Economist, CAIA, Data Scientist, Quant, and Computer Scientist who channels her interdisciplinary applied sciences background in the intersection of financial markets and technology. Sol is the CEO and founder of Data Innovation Labs, a full-service Data Science, and Decision Intelligence consulting group deploying 4IR technologies, Ai and implementing digital transformation across business verticals.

Sol’s extensive education spans top higher learning institutions from the Pontificia Universidad Católica of Chile (Commercial Engineering, Macroeconomics, Econometrics, Finance, Management), Indiana University — Kelly School of Business (MBA specializing in Derivative Securities), and Harvard University (Computer Science concentration Data Science, and Financial Crisis).

Sol graduated top of her class from Harvard University and she holds an Academic Teaching Fellow positions for Data Science at Harvard. She has also judged the University of Chicago Financial Mathematics Master Program Graduating Competition.

Her financial career spans almost 3 decades with domain knowledge in alternative investments. Sol has traded, modeled, and developed trading strategies for the universe of securities and contracts — including derivatives. She has been one of the first and few quants worldwide who have successfully modeled a wide spectrum of agrarian commodities in the Hedge Fund space, and she was one of the first Data Scientists and to embrace Blockchain and DLT technologies to leverage AI and Deep learning, engaging her in cutting edge international collaborations.

Her training as a Data Scientist and her deep knowledge of AI helps her in learning how to overcome her unfortunate new hearing disabilities. For that, she found motivation in how Artificial Neural Networks learn, and she is successfully defying severe nuero-sensorial hearing disabilities for which there’s no help with the current state of medical advancements. By leveraging the theory and math of AI she is able to achieve inclusion as a leader, professional, and inspire.

Reference:

• Anderson RM, Fraser F et al. Epidemiology, transmission dynamics and control of SARS: the 2002–2003 epidemic. Phil Trans Roy Soc Ser B 2004; 359: 1091–490.
• Asia Times. China suppressed CoViD-19 with AI and Big Data. https://asiatimes.com/2020/03/china-suppressed-covid-19-with-ai-and-big-data/
• Cauchemez S, Ferguson NM, Wachtel C, et al. Closure of schools during an influenza pandemic. Lancet Infect Dis 2009; 9: 473–81.
• Chinazzi, Matteo et al. Estimating the risk of sustained community transmission of COVID-19 outside of Mainland China. medRxiv 2020.
• Chinese Center for Disease Control and Prevention. Distribution of new coronavirus pneumonia. http://2019ncov.chinacdc.cn/2019-nCoV/
• Elston JW, Cartwright C, Ndumbi P, Wright J. The health impact of the 2014–15 Ebola outbreak. Public Health 2017; 143: 60–70.
• European Centre for Disease Prevention and Control. Daily risk assessment on COVID-19. 2020. https://www.ecdc.europa.eu/en/current-risk- assessment-novel-coronavirus-situation.
• Fraser C, Riley S, Anderson RM, Ferguson NM. Factors that make an infectious disease outbreak controllable. Proc Natl Acad Sci USA 2004; 101: 6146–51.
• Google DeepMind. https://deepmind.com/research/open-source/computational-predictions-of-protein-structures-associated-with-COVID-19.
• Imperial College London, MRC Centre for Global Infectious Disease Analysis. News / COVID-19 — report 3: transmissibility of 2019-nCoV. 2020.
  Feb 21, 2010. https://www.imperial.ac.uk/mrc-global-infectious-disease- analysis/news — wuhan-coronavirus/.
• Hellewell J, Abbott S, Gimma A, et al. Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts. Lancet Glob Health 2020; published online Feb 28. https://doi.org/10.1016/S2214-109X(20)30074-7.
• Hollingsworth TD, Klinkenberg D, Heesterbeek H, Anderson RM. Mitigation strategies for pandemic influenza A: balancing conflicting policy objectives. PLoS Comput Biol 2011; 7: e1001076.
• Hu, Zixin, et al. Artificial Intelligence Forecasting of Covid-19 in China. medRxiv 2020.
• Keeling MJ, Hollingsworth TD, Read JM. The efficacy of contact tracing for the containment of the 2019 novel coronavirus (COVID-19). medRxiv 2020; published online Feb 17. https://doi.org/10.1101/2020.02.14.20023036 (preprint).
• Li L, Wong JY, Wu P, Bond HS, Lau EH, Sullivan SG, Cowling BJ. Heterogeneity in estimates of the impact of influenza on population mortality: a systematic review. Am J Epidemiol 2018; 187: 378–88.
• Li Q, Guan X, Wu P, et al. Early transmission dynamics in Wuhan, China, of novel coronavirus-infected pneumonia. N Engl J Med 2020; published online Jan 29. DOI:10.1056/NEJMoa2001316.
• Lipsitch M. Will COVID-19 go away on its own in warmer weather? Center for Communicable Disease Dynamics (CCDD) at the Harvard T.H. Chan School of Public Health. 2020. https://ccdd.hsph.harvard.edu/ will-covid-19-go-away-on-its-own-in-warmer-weather/.
• Lloyd-Smith JO, Schreiber SJ, Kopp PE, Getz WM. Superspreading and the effect of individual variation on disease emergence. Nature 2005; 438: 355–59.
• Makoni M. Africa prepares for coronavirus. Lancet 2020; 395:483.
• New York Times. North Korea Bans Foreign Tourists Over Coronavirus, Tour Operator Says. https://www.nytimes.com/2020/01/21/world/asia/ coronavirus-china-north-korea-tourism-ban.html, 2020.
• Pinotti, F., et al. Lessons learnt from 288 COVID-19 international cases: importations over time, effect of interventions, underdetection of imported cases. https://www.epicx-lab.com/uploads/9/6/9/4/9694133/inserm_covid-19_ imported-cases_20200223.pdf, 2020.
• Rambaut, A. Phylogenetic analysis of nCoV-2019 genomes. http://virological. org/t/phylogenetic-analysis-of-23-ncov-2019-genomes-2020–01–23/335, 2020.
• Reuters. Russia ramps up controls, shuts China border crossings over virus
  fears. https://www.reuters.com/article/us-china-health-russia-border/ russian-regions-in-far-east-close-border-with-china-amid-coronavirus-fears-tass-idU 2020.
• The Australian. ”Travelers from China to be denied entry to Australia. https://www.theaustralian.com.au/subscribe/news/1/?sourceCode=TAWEB_WRE170_a&dest=https%3A%2F%2Fwww.theaustralian.com.au%2Fnation% 2Ftravellers-from-china-to-be-denied-entry-into-australia%2Fnews-story% 2F7b7619d44af78dd7395a934e22b52997&memtype=anonymous&mode=premium, 2020.
• The Washington Post. Why outbreaks like coronavirus spread exponentially, and how to “flatten the curve”.https://www.washingtonpost.com/graphics/2020/world/corona-simulator/
• The Novel Coronavirus Pneumonia Emergency Response Epidemiology Team. The epidemiological characteristics of an outbreak of 2019 novel coronavirus disease (COVD-19). China CDC Weekly 2020; 2: 113–22.
• Toui tre News. Vietnam aviation authority ceases all flights to and from coronavirus-stricken Wuhan. https://tuoitrenews.vn/news/business/20200124/ vietnam-aviation-authority-ceases-all-flights-to-and-from-coronavirusstricken-wuhan/ 52707.html, 2020.
• UK Government, Department of Health and Social Care. Coronavirus action plan. March 3, 2020. https://www.gov.uk/government/publications/coronavirus-action-plan.
• Wand, Shuai, et al. A deep learning algorithm using CT images to screen for Corona Virus Disease (COVID-19). https://doi.org/10.1101/2020.02.14.20023028.
• WHO. Coronavirus disease 2019 (COVID-19) situation report — 44.
  March 4, 2020. https://www.who.int/docs/default-source/coronaviruse/ situation-reports/20200304-sitrep-44-covid-19.pdf?sfvrsn=783b4c9d_2.
• WHO. Coronavirus disease (COVID-2019) situation report — 30. https://www.who.int/docs/default-source/coronaviruse/situation- reports/20200219-sitrep-30-covid-19.pdf?sfvrsn=3346b04f_2.
• Yang, Zifeng, et al. Modified SEIR and AI prediction of the epidemics trend of COVID-19 in China under public health interventions. Journal of Thoracic Disease. 2020. http://dx.doi.org/10.21037/jtd.2020.02.64
• Yu M, He S, Wu D, et al. Examining the multi-scalar unevenness of high-quality healthcare resources distribution in China. Int J Envois Res Public Health 2019; 16: e2813.
• Zhang, Q., K. Sun, M. Chinazzi, A. Pastore y Piontti, N. E. Dean, D. P. Rojas, S. Merler, D. Mistry, P. Poletti, L. Rossi, M. Bray, M. E. Halloran, I. M. Longini, and A. Vespig- nani. Spread of Zika virus in the Americas. Proceedings of the National Academy of Sciences, 114(22):E4334–E4343, 2017.
• Zou L, Ruan F, Huang M, et al. SARS-CoV-2 viral load in upper respiratory specimens of infected patients. N Engl J Med 2020; published online Feb 19. DOI:10.1056/NEJMc2001737.
• https://www.fiercehealthcare.com/tech/boston-startup-using-ai-remote-monitoring-to-fight-coronavirus