coronasafe / analytics Goto Github PK

Context:

The War Room in Ernakulam needs advanced data based pandemic management predictions for Ernakulam administration to plan an unlock phase and more importantly remain unlocked in order for public economic activity to move forward into a continuum.

At present, the government or public are not basing their day to day decisions based on Oxygen bed availability, which is the threshold of capacity that the district can manage.

During the third peak in May of 62,607 patients (see figure 1) the healthcare infrastructure systems were stretched to the limit at EKM and full lockdown implemented.

Desired Solution:

A real-time data analytics can enable district administration to plan future scenarios of how various unlock scenarios could be implemented, thus leading to various levels of freedom to go about with economic activity.

From the offline analysis of 14 months of unlock scenarios from March 2020 till May 2021, the economic activity allowed by the administration can be grouped into five levels. The administration has created 11 different lockdown stages across the last 428 days. We never reached Level 5 (Schools Re-opening)

The chart below shows the various levels of activity allowed during the unlock phase with the maximum phase being level four before a large third wave with the Delta variant pushing the healthcare capacity to limit forcing a full lockdown of 14 days in May 2021.

Figure 2 - Levels of Economic Activity Achieved during Unlock

Forward Planning:

The model should predict future bed usage by using past data of how many people per 10,000 needed o2 bed and approximately for how many days. The growth of the virus needs to be remodelled from the infected numbers and R value calculated for every month and how the spread changed across various lockdown stages.

This time-series R value should be matched to how fast the beds are being used up. Finally, the ability to manage the patient load at a particular point in time with a safety parameter of 95% utilisation shall indicate what level of economic activities are permissible.

The Goal is to keep vaccinating more people and continue to progressively unlock from Level 1 to Level 5. Level 5 being the time when schools are open and life is back to near pre-pandemic normalcy.

For example: we have 5000 oxygen beds in Kochi.

Above 1000 free o2 beds: step 1
Above 2000 free o2 beds: step 2
Above 3000 free o2 beds: step 3
Above 4000 free o2 beds: step 4

The more free capacity we have, the more freedom that we can give to citizens to carry out economic activities without congregating as groups.

The incoming data should ingested to finetune the model in real time and by giving the ability to a public health experts to test various scenarios by changing few variable such as R value, the system should be able to suggest when we can safely move to a level higher or advice if we should move a step down as precaution to rising bed occupancy.

This prediction should also include overall factors of EKM like

Total population : 32,000,000
Total Vaccinated with both:248,133
Total Vaccinated with one shot: 1,202,650

The efficiency of the AstraZeneca shots is 33% after the first shot and 60% after second.. Thus, 396,874 people have immunity from first and 163,768 people from both shots have immunity. The total tested positive and discharged form our Care Database as on (June 11th) are 195,977 taking people with vaccine+immunity as 756,619 (195977+396874+163768)

The Public Health Experts in Kerala are assuming we catch only 1 in 4 of positives, we have 235,696*4 = 942,784 infected who may have acquired immunity.

Total Safe Population to Virus (Vaccine+Acquired Immunity = 756,619+942,784 = 1,699,403
Total Remaining to get infected = (32,000,000 - 1,699,403)= 1,500,597 (46%)

At the last peak of 60,000 patients in the month of May, Ekm hit its limits and had to lockdown for two weeks for things to calm down. With 46% or 15 Lac people yet to get infected, a situation where the overall infection is 60,000 will overburden the system.

However, within that limit, we can safely continue our activities. This means that with 30,000 cases or 40,000 cases per month which translates to 1000 cases per day, EKM could possibly continue to keep the economy open if R value is X?

We need to calculate this X so that we know that the surge will not climb to intolerable levels.

Once we have the ability to publish this data dynamically of what the free feds are and what level of unlock we are in, people may be able to carry out their own personal vigil to ensure that mass gatherings and too much mingling are not happening to increase the value to a situation where there has to be another higher level of lockdown.

For common freedoms to increase, we need to act together as a society and defeat the pandemic with more vaccination and increased social vigil that is enforced where needed with a strong iron fist, if some small section is not cooperating with the administration.

This modeling capability can be continuously tweaked with more granular data that includes seeing the detailed census data of EKM (that is available) and can be interpreted with male and female and how much did in a particular age group, the virus had an attack vector leading to oxygen bed usage.

We have age group-wise census data from 2011 as well as age group-wise hospitalization in care that can be used. This sort of granular data would further fine tune the system.

Once the administration understands the value of the tool, the data collection and feed into the system would improve creating more accurate models of prediction.

Eventually, we realize that the model and how people act on the field would be different.

Thus, even if people don't cooperate, the increase in bed usage should give the administration a good heads up warning of when the system would breach the threshold capacity of 5000 beds during a surge and thus by when to activate a higher level of lockdowns, if spread is not reducing.