Fighting COVID19 - Granular Data and Effective Reporting is Must

To fight COVID19 we need quality data at the highest level of granularity. This data must not only be accurate but must enable us to view the trend. As only then we can conclude where more intervention is required (epidemic is going out of control) and where the lockdown can be relaxed (the situation is improving) - an absolute must due to huge suffering largely caused to the poor and homeless. Unfortunately, the information that we are playing with (at the public level) is only granular at the state level. Next level information (district and city) is almost missing. Even if the same is available, it is not in the form that can be used for making any worthwhile predictions (at least the data available in the public domain). For example, the information available at the website of the ministry of health and family welfare is aggregated information. It does not give a time series. It does report the district level information, but with a caveat - "Likely to change once districts of all known 6761 cases are ascertained". Also, we have no idea whether this information is updated and current. This means we can't even rely on this information for understanding the extent of the spread. Leave aside the intensity. Hence, I attempted to extract this information to the best possible and packaged it in a manner that can be put to some use by those interested.

India COVID-19 Tracker

For tracking the information at the next level of granularity I am using this site (link embedded in the heading). As of all the sources that I have researched, this site has the best information. But unfortunately, it looks like things have broken down here as well. Their original intent was to track COVID-19 patients by patient number (raw data page), but I find this ID missing in all the other pages it maintains. This site also reports information at the district level, but the time series is missing. Hence no trend analysis is possible. With no recourse, I used the "raw data page" to extract this trend. The analysis based on this information, I have posted on my site which I have been maintaining for some other purpose. These pages are titled COVID19- India Cases, and COVID19- India District Cases. One can see the treads here, but the accuracy is totally dependent upon the quality of the data posted. Please note that some cases are not even mapped to their respective districts (total case and those linked with districts do not match). For such information, I have tagged them to their respective states for the purpose of reporting and analysis.

Analysis

Using this data source I have created some charts which help us to dive into the next level of granularity, i.e at the district level. Ideally, we would have liked this at the city level, but we have to be happy with what we have. Using this information I have identified the districts with reported cases in a time series. Enabling us to see the extent of the spread and analyze the trends. These charts are embedded under.

On can clearly see that more than 50% of the districts are affected. 281 new districts got added after the lockdown was announced on 22nd March 2020. This is 3 times the number that was affected at the time of the lockdown. Using the above charts and the one that I have posted on my website COVID19- India District Cases, one can do trend analysis and see in which districts the cases are slowing down.

The trend of the past 15 days for the top 25 districts (output of the tool) is as under.

The tool empowers the user to increase the number of districts and days in view. It can also sort them using various matrices as defined in the tool.

Conclusion

My effort is more amateurish than anything close to a professional level. The attempt is to evaluate the direction where we all need to go. Maybe the govt is already doing this, but we are not sure. As we do see the information at the public level pointing in this direction. This is a must of enabling effective collaboration and management. Hence, Apart from all other good work that is going on in the field, we need to also focus on aggregating and reporting the best quality data on COVID19 patients at the highest level of granularity. This alone will help us direct our efforts with utmost efficiency and with surgical precision - to save both lives and livelihood - must for the poor who are the worst impacted by this pandemic.

(Views expressed are of my own and do not reflect that of my employer)

PS: Find the list of other relevant articles in the embedded link.