Introduction
Related works
Susceptible-exposed-infected-death (SEIRD) epidemic model
Machine learning approaches
Issues in vaccine production and supply
Materials and methods
Data
-
We consider 5 towns, called boroughs, in New York City (NYC), viz. Bronx, Brooklyn, Manhattan, Queens and Staten Island. We obtain the borough data, such as Gross Domestic Product (GDP), population density, etc., from Wikipedia (Neighborhoods 2020). COVID infection and deaths are taken from The City (Coronavirus 2020) based on records of Department of Health and Mental Hygiene. We use NYC Health records (Nyc health 2020) that show daily infected from March-August 2020 from New York Department of Health.
-
We source the mobility data of NYC traffic from NYCOpenData (Nycopendata 2020)—a repository for fields ranging from city government, education, environment, health to public safety, recreation, social services and transportation. The stated data (spanning a period from 2014 to 2019), collected by the Department of Transportation of New York Metropolitan Transportation Council (NYMTC), has the following fields: ID, road name, source and destination intersecting street name, compass direction, date and time. We calculate the transition matrix (see “Inter-zone mobility model” section) that captures the probability of travelling within and across boroughs.
SEIRD epidemic model
Ordinary differential equations
Scenario
Inter-zone mobility model
Minimization of Kullback–Leibler divergence
Modeling hospitalization queue
Reinforcement Q-learning model (RL)
Action space
Reward function
Parameter | Notation | Value |
---|---|---|
Number of iterations | – | 100 |
Simulation duration | T | 180 days |
Number of boroughs | |B| | 5 |
Number of lockdown levels | \(\chi\) | 6 |
Interval for invoking ODE | \(\eta\) | 12 hours |
SEIRD parameters(Korolev 2020) | \(\sigma , \gamma , \alpha\) | 0.25, 0.1, 0.05 |
Interval and initial number of new infection | \(\zeta , k\) | 30 days, 200 |
Migration rate | \(\zeta\) | 0.01 |
Treatment rate | r | 0.0029 |
Hospital fatality rate | \(\alpha _h\) | 0.2 |
SEIRD infection probability | p | 0.01 |
Collision diameter | d | 1 m |
RL probability of random action | ep | 0.75 |
RL decay factor | dc | 0.99 |
RL transition window | W | 5.5 days |
Number of probability of queue levels | \(\omega\) | 3 |
Levels in probability of queue | \(w_1, w_2, w_3\) | \(0 - 0.33\), \(0.33 - 0.66\), \(0.66 - 1.0\) |