Thread by @mohdhelmi

The written response to parliamentary questions of MySejahtera tracing rate of 4% in the parliament made a headline. I am fearful of erosion of public trust.

I want to share some of my analysis to recover the image of e-tracing solution.

tinyurl.com/yy4423uc

Let me be clear that this is not a condemnation or trying to "outdo" others. Not Selangkah VS MySJ.

Au contraire, I hope this thread returns the public confidence to e-tracing.

It matters less which one choose to scan, but the bigger risk if people forgo scanning altogether.

As a proponent of e-tracing and the pioneer for QR tracing solution, I believe the 4% figure was starkly underestimated.

I'd like to share our analysis of @SELangkah_cvd19 Contact Tracing efficiency during its peak performance.

Our Contact Tracing efficiency of Selangkah (May 5,2020 - September 30, 2020 ).

Why we choose May 5? It was when @SELangkah_cvd19 was born. As the 1st QR solution, we were finetuning.

1) On the stipulated period: A record of 4871 people is diagnosed with Covid19 nationwide.

2) A Selangkah retrospective analysis of its datalake found 3855 corresponding phone numbers of patients in (1) existed in its datalake.

[How we get patient's phone number? We simply ask them directly. Did we made it public ?, you know we dont.]

3) Out of the number in (2) , 2757 phone numbers has "interaction" with each others, meaning that they could have been traced earlier in Selangkah".

The balance 1098 is 'isolated' (exist in Selangkah, but no interaction/cannot be linked with other patients' phone numbers).

4) Hence Selangkah algorithm were able to "Patient - suspect" pair for Contact Tracing at the rate of : 2757 / 4871 * 100 = 56 %

* this is an underestimate. The penetration and use of Selangkah decrease in the premise after MySJ is made mandatory by Aug 8.

Despite this significant drop of use, Selangkah / any e-tracing solution would have been able to trace 56% of cases happening in the population on the stipulated period.

I believe it CAN DO BETTER WITH MORE PREMISES AND USER PARTICIPATING.

In e-Contact Tracing - its not only about how big the datalake is, or how wide the penetration is.

It's about how efficient the algorithm in pinpointing potential cases among the sea of "noises". It's finding needles in a haystack.

I believe MySejahtera figure must be higher / at least closer to @SELangkah_cvd19

Their penetration is bigger, and usage higher.

The 2 solutions might have differences in "pairing & dwell-time algorithm", but I dont believe that it would be drastically that low, at 4%.

I hope this calm the public and everyone will continue to opt for a more efficient solution, that is e-tracing compared to manual paper record.

Don't do it because of the RM1k compound, do it to help the frontliners - once again.
[end]
#kitajagakita

unroll @threadreaderapp pls