A Schematic of DOCL project
Written by Seoyeon Kim, Seunga Oh, Strategy Team
Edited & Revised by Deokjae Han, MD, CMO
Translated by Derek Wonwoo Sohn, IMS
(International Marketing Supervisior)
At the beginning of this new decade, the world has confronted an unprecedented proliferation of the life-threatening virus COVID-19 (SARS-CoV-2, severe acute respiratory syndrome–Coronavirus-2). Since then the world has entered ‘the era of infection’ and is going to witness the virus periodically visiting and haunting the global society for years to come. Many leading scientists and scholars have already proclaimed ‘war’ with the virus, and thus have warned that humanity should be proactively prepared for ‘post COVID-19.’
At the center of efforts for preventive measure is software platform equipped with AI and machine learning.
The power of AI & ML-based prediction model has already been demonstrated by the Canada-based AI-platform, BlueDot, which detected the global expansion of COVID-19 earlier than WHO (World Health Organization) did. BlueDot is the AI platform designed by a group of doctors to prevent the like incident of SARS (SARS-CoV, severe acute respiratory syndrome–Coronavirus) that caused another disaster back in 2003. Last year, the doctors published the first scientific paper on spread of COVID-19; in this paper , the BlueDot correctly predicted that COVID-19 would spread from Wuhan to Seoul, Tokyo, and Hong Kong, 9 days ahead of WHO’s statement. This is an exemplary use case of AI-based forecast model.
Importance of data collection
Meanwhile, patient data will be an ammunition to fight against the war and a great instrument that can help predict the spread of virus. The patient data can enhance the efficiency in severity classification of patients on critical care and thus in the resource allocation of doctors and hospitals. Such change can eventually increase the likelihood of saving the lives of high-risk patients, many of whom had lost theirs during the chaotic period of early COVID-19 pandemic.
According to the research paper registered in SJTREM (Scandinavian journal of trauma, resuscitation and emergency medicine) , the emergency medical staff was able to apply the basic information of emergency patients to AI algorithm and then diagnose more accurately which patients needed critical care than could existing emergency triage models. While context may differ between circumstances of EMS (emergency medical service) and of infectious disease, it is meaningful to realize the ability that AI algorithm—in combination with the collected patient data—can exert on agile prioritization of patients treatment in accordance with the criticality of their symptoms.
Difficulties in data collection
Despite knowing how important accumulating the patient data is, medical staff still faces difficulties in applying the given data to their clinical studies at the right time. Unfortunately, the delay is unexpectedly long from case collection to its meaningful application.
According to the research by IPH, Institute of Public Health, it takes as long as 17 years for the medical researchers to incorporate the up-to-date research data into actual treatment at the point-of-care environment. To infectious disease research whose shareability is at its utmost importance, this lead time and its time variation are critical barriers .
DOCL, to overcome
Then, how can doctors create a virtuous cycle of data collection, insight generation and strategy winning in this COVID-19 pandemic situation? DOCL (Doctors on the cloud) is the project designed to create this cycle and further facilitate it with the AI-based triage and prognosis prediction model. This project is intended to streamline the information delivery process from the breakout of a disease to application of clinical findings to the hospitals.
DOCL team endeavors to use the ML training to turn the day-to-day bits of clinical data into meaningful information and advance for more accurate prediction and ultimately for effective disease prevention.
The methodology is as follows.
- Patient screening based on the verified symptoms of COVID-19 in line with up-to-date research findings
Diverse case studies and meta analyses of research paper allowed us to spot the patient symptoms that are likely to occur and to structure the menu and usage flow of our patient app.
- Patient-to-doctor communication channel and data sharing
DOCL team created two interconnected apps, one for patients and another for doctors. If patients use smartphone or an internet browser of any device to record their symptoms, then the patients will be shown the level of their necessity for COVID-19 testing or hospitalization. Their written records are saved to the DOCL’s database.
To tightly manage the anonymity of patients, DOCL asks the only private information to patients, the email address. Since randomly generated codes replace the email addresses and are synced into medical staff app, the patients’ privacy is secured. The further information that patients record is simultaneously sent to medical staff. In case of inpatients, a doctor can input additional data such as test results, vital signs, medicine into the medical staff app at any time. Moreover, a doctor can also view within the app the information recorded by self-quarantined, discharged patients, or outpatients and respond to patients based on the given information.
- Machine Learning and system updates
The input data by patients and medical staffs are analyzed real time thanks to the capability of machine learning. Because the most recent clinical data are updated, the prediction result that reflects the real-time learning from the new data is adjusted and delivered to patients.
Significance of DOCL
The significance of DOCL project is as follows:
- As the machine learning continues to absorb and learn recent and unprecedented data of its severity and prognosis of COVID-19, it will enhance the accuracy level of prediction as well as the overall medical treatment quality.
- The app-to-app connection between doctors and patients expediate the real-time observation and introduction of preventive measures before patients’ condition worsens.
- The reduction of unnecessary contact between a patient and medical staff leads to lower chance of getting infected by disease.
With your help, we can save the world
However, to fully achieve these stated values, we need to be able to collect the most effective and accurate data. In short, this success depends on the active participation by doctors, nurses and medical personnel. While AI will take care of its part to create algorithm and categorize patient triage, data collection and its quality assurance, which empower AI to do its job, are in the hands of us, the people in the medical field.
The ultimate goal of this not-for-profit project is to develop a competitive model and distribute this product to developing countries where medical resource is severely deficient. DOCL team saw this as responsibility and opportunity for South Korea to help solve the riddle of the COVID-19 pandemic problems.
The DOCL project will be a cornerstone to that end; its initiative will amass the international cooperation and common good of all society.
Thus, we ask for your participation into this endeavor. Active participation by doctors raises the model’s accuracy and confidence level, attract more patients for their diagnosis and eventually help create a healthy ecosystem of critical medical information sharing. When the next pandemic that threatens humanity sweeps the world, we hope to base this project as a platform to forecast ahead of time and help prevent the disease much more effectively .
About Project and Team
DOCL is a non-profitmaking project initiated by six medical specialists (army surgeons) currently serving military duty at Armed Forces Medical Command and Armed Forces Capital Hospital under Ministry of Defense, Rep. of Korea. Facing the COVID-19 crisis, these young doctors decided to help solve the existing problems of medical procedures regarding COVID-19 by developing prediction model of disease, severity classification, and efficient information sharing. The project kicked off with Yonsei University College of Medicine signing MOU, and soon the project was registered to WHO Digital Atlas after having developed the early version of prediction model. Since the project took off successfully, DOCL team has been collaborating with the government agencies such as Ministry of Defense, Ministry of Foreign Affairs, and Korea National Institute of Health under Ministry of Health and Welfare. DOCL is also working in close cooperation with Seoul metropolitan government, and as a result, the quantity and quality of data analysis improved significantly. The only goal of DOCL project is to make the prediction model work and prevent disease so that it can provide welfare to patients as well as safety and convenience to medical staff based not only in Korea but in other countries.
 I.I. Bogoch, et al. Pneumonia of Unknown Aetiology in Wuhan, China: Potential for International Spread via Commercial Air Travel. J Travel Med). 2020 Mar 13;27(2).
 D. Kang, et al. Artificial intelligence algorithm to predict the need for critical care in prehospital emergency medical services. SJTREM. 2020. 28(17)
 Z.S. Morris, et al. The answer is 17 years, what is the question: understanding time lags in translational research. SAGE journals. 2011. 104(12). p.510-520
 A.Alimadadi, et al. Physiol Genomics. 2020 Apr 1; 52(4): 200–202. Published online 2020 Mar 27. doi: 10.1152/physiolgenomics.00029.2020