Leo Bastos
@leosbastos.bsky.social
Public health researcher at Fiocruz 🇧🇷
Associate professor at IMPA Tech 🇧🇷
Bayesian statistician 📈
Infectious diseases modelling 🦠🦟
Father of 3 boys (be nice).
Associate professor at IMPA Tech 🇧🇷
Bayesian statistician 📈
Infectious diseases modelling 🦠🦟
Father of 3 boys (be nice).
This way I don't forget mentioning anyone. :-) Thanks everyone.
September 4, 2025 at 6:37 PM
This way I don't forget mentioning anyone. :-) Thanks everyone.
Here one can see how heterogeneous is the relationship between the observed dengue cases and the model predictions in country by health regions in the two seasons (2022-2023 and 2023-2024).
September 4, 2025 at 6:37 PM
Here one can see how heterogeneous is the relationship between the observed dengue cases and the model predictions in country by health regions in the two seasons (2022-2023 and 2023-2024).
We suggest the use of posterior predictive quantiles to define epidemic levels.
September 4, 2025 at 6:37 PM
We suggest the use of posterior predictive quantiles to define epidemic levels.
NEW PAPER
We developed a statistical model to forecast dengue cases for the next season using the history of cases only. We suggest using the posterior predictive to define epidemics bands.
For instance, we could see how huge was 2024 dengue season in Brazil.
www.sciencedirect.com/science/arti...
We developed a statistical model to forecast dengue cases for the next season using the history of cases only. We suggest using the posterior predictive to define epidemics bands.
For instance, we could see how huge was 2024 dengue season in Brazil.
www.sciencedirect.com/science/arti...
September 4, 2025 at 6:37 PM
NEW PAPER
We developed a statistical model to forecast dengue cases for the next season using the history of cases only. We suggest using the posterior predictive to define epidemics bands.
For instance, we could see how huge was 2024 dengue season in Brazil.
www.sciencedirect.com/science/arti...
We developed a statistical model to forecast dengue cases for the next season using the history of cases only. We suggest using the posterior predictive to define epidemics bands.
For instance, we could see how huge was 2024 dengue season in Brazil.
www.sciencedirect.com/science/arti...
I usually don't like pictures of myself, but I like this one.
August 9, 2025 at 11:31 PM
I usually don't like pictures of myself, but I like this one.
When Thomas Bayes meets John Snow. Presented in Vitória, capital of Espírito Santo state, at the 69th RBRAS meeting and 21st SEAGRO
August 5, 2025 at 4:07 AM
When Thomas Bayes meets John Snow. Presented in Vitória, capital of Espírito Santo state, at the 69th RBRAS meeting and 21st SEAGRO
We had a very special visitor today, @paulamoraga.bsky.social gave us a great lecturer on Spatial data science in R.
Thanks Paula!
Thanks Paula!
June 26, 2025 at 9:46 PM
We had a very special visitor today, @paulamoraga.bsky.social gave us a great lecturer on Spatial data science in R.
Thanks Paula!
Thanks Paula!
My team (Cruzeiro 🦊) is not there in the World Cup, but it is nice to see the Brazilian clubs playing well at there. It is good for me since we are leading* the Brazilian league (Brasileirão).
*Second place, same number of points as Flamengo, but they scored more goals.
*Second place, same number of points as Flamengo, but they scored more goals.
June 21, 2025 at 9:01 PM
My team (Cruzeiro 🦊) is not there in the World Cup, but it is nice to see the Brazilian clubs playing well at there. It is good for me since we are leading* the Brazilian league (Brasileirão).
*Second place, same number of points as Flamengo, but they scored more goals.
*Second place, same number of points as Flamengo, but they scored more goals.
Today (25/5) is Fiocruz 125th year anniversary, instead of the iconic castle picture (which turns out I have no one in my phone) I post a picture of the cell from the Life's museum which is part of the Fiocruz campus in Manguinhos and I place I often take my kids to play around there.
May 25, 2025 at 2:47 PM
Today (25/5) is Fiocruz 125th year anniversary, instead of the iconic castle picture (which turns out I have no one in my phone) I post a picture of the cell from the Life's museum which is part of the Fiocruz campus in Manguinhos and I place I often take my kids to play around there.
Since 2010, the curve of cases this year is the second worst. The worst one was last year.
info.dengue.mat.br/alerta/35503...
info.dengue.mat.br/alerta/35503...
May 2, 2025 at 6:58 PM
Since 2010, the curve of cases this year is the second worst. The worst one was last year.
info.dengue.mat.br/alerta/35503...
info.dengue.mat.br/alerta/35503...
Dengue in the city of São Paulo, Brazil. Compared to last year, this year doesn't look that bad. But...
May 2, 2025 at 6:58 PM
Dengue in the city of São Paulo, Brazil. Compared to last year, this year doesn't look that bad. But...
In this plot, we have the last week nowcasting cases over this week's dataset release. Since the delay is not too big, the correction worked quite well with few time series.
April 4, 2025 at 7:45 PM
In this plot, we have the last week nowcasting cases over this week's dataset release. Since the delay is not too big, the correction worked quite well with few time series.
Quick update on notification delay correction (nowcast) of measles cases in the US after the new update from the CDC website. The number of cases are still going up.
April 4, 2025 at 7:45 PM
Quick update on notification delay correction (nowcast) of measles cases in the US after the new update from the CDC website. The number of cases are still going up.
After organizing the data, here is the result applying a simple Bayesian chain-ladder model. The number of cases are not reducing, on the opposite the cases are still increasing.
March 29, 2025 at 2:17 PM
After organizing the data, here is the result applying a simple Bayesian chain-ladder model. The number of cases are not reducing, on the opposite the cases are still increasing.
The delay problem, since the CDC decided to report cases in a regular pattern (every Friday) we can save all time series and after a couple of them we are able to apply a statistical model to correct the delays
March 29, 2025 at 2:17 PM
The delay problem, since the CDC decided to report cases in a regular pattern (every Friday) we can save all time series and after a couple of them we are able to apply a statistical model to correct the delays
Since the CDC was updating the time series of number of reported measles cases in their website I decided to try out the delay-correction because the big drop in the time series is just reporting delaying. (It is a well-know problem in epidemiological surveillance)
March 29, 2025 at 2:17 PM
Since the CDC was updating the time series of number of reported measles cases in their website I decided to try out the delay-correction because the big drop in the time series is just reporting delaying. (It is a well-know problem in epidemiological surveillance)
While putting the kids to sleep it seems Shakira is testing* the sound in the Stadium for her gig on the 11th.
* I can hear her voice but it might be a recording.
* I can hear her voice but it might be a recording.
February 9, 2025 at 1:57 AM
While putting the kids to sleep it seems Shakira is testing* the sound in the Stadium for her gig on the 11th.
* I can hear her voice but it might be a recording.
* I can hear her voice but it might be a recording.
In Brazil, the main printed journals (Folha de São Paulo and O Globo) also avoid naming the fascist style salute.
January 21, 2025 at 3:08 AM
In Brazil, the main printed journals (Folha de São Paulo and O Globo) also avoid naming the fascist style salute.
The notification system is not the best place to look at deaths (deaths should be reported in another system the mortality one). But still, I would assume the mortality due to snake bites shouldn't be big.
Here is a table with cases per year (rows) and case evolution (columns).
Here is a table with cases per year (rows) and case evolution (columns).
January 7, 2025 at 10:09 AM
The notification system is not the best place to look at deaths (deaths should be reported in another system the mortality one). But still, I would assume the mortality due to snake bites shouldn't be big.
Here is a table with cases per year (rows) and case evolution (columns).
Here is a table with cases per year (rows) and case evolution (columns).
I will add an interesting case in Brazil. The city of Joinville, located in the South region of Brazil where dengue wasn't a reality.
From 2010 to 2019, there is no record of dengue outbreaks. From 2019 onwards, it is an impressive sequence of recording breaking outbreaks.
From 2010 to 2019, there is no record of dengue outbreaks. From 2019 onwards, it is an impressive sequence of recording breaking outbreaks.
November 10, 2024 at 2:13 AM
I will add an interesting case in Brazil. The city of Joinville, located in the South region of Brazil where dengue wasn't a reality.
From 2010 to 2019, there is no record of dengue outbreaks. From 2019 onwards, it is an impressive sequence of recording breaking outbreaks.
From 2010 to 2019, there is no record of dengue outbreaks. From 2019 onwards, it is an impressive sequence of recording breaking outbreaks.
For very selfish reasons I am looking at MG (my birth state) and RJ (where I live). The regions in MG where my parents came from (east and northeast of the state) have higher rates of ppl living over 100, so I hope there is a genetic component. Rio could be a bit better though.
November 3, 2024 at 3:53 AM
For very selfish reasons I am looking at MG (my birth state) and RJ (where I live). The regions in MG where my parents came from (east and northeast of the state) have higher rates of ppl living over 100, so I hope there is a genetic component. Rio could be a bit better though.
I have watched the Netflix doc "Live to 100: Secrets of the Blue Zones", so I decided to look at the official stats and calculate the rate of centenaries per 10.000 inhabitants by municipality in Brazil. I am quite impressed with the Northeast. Most cities with high rates are very small.
November 3, 2024 at 3:41 AM
I have watched the Netflix doc "Live to 100: Secrets of the Blue Zones", so I decided to look at the official stats and calculate the rate of centenaries per 10.000 inhabitants by municipality in Brazil. I am quite impressed with the Northeast. Most cities with high rates are very small.
And I did though about using an app or something to help me out, but it is always important to remind you that
November 2, 2024 at 7:23 PM
And I did though about using an app or something to help me out, but it is always important to remind you that
Today I will give a talk at the National School of Statistical Sciences (ENCE/IBGE) on how my groups tackled different problems related to the COVID-19 epidemics in Brazil.
It was interesting building the slides mixing the COVID-19 events with our group publications.
It was interesting building the slides mixing the COVID-19 events with our group publications.
October 29, 2024 at 12:35 PM
Today I will give a talk at the National School of Statistical Sciences (ENCE/IBGE) on how my groups tackled different problems related to the COVID-19 epidemics in Brazil.
It was interesting building the slides mixing the COVID-19 events with our group publications.
It was interesting building the slides mixing the COVID-19 events with our group publications.
It is grant writing season (at least for me). My projects are essentially adapting Bayesian statistical methods in infectious disease epidemiology problems. But it seems to me that in order to make them competitive I need to add fancy words like learning and artificial intelligence...
#StatsSky
#StatsSky
October 21, 2024 at 8:54 AM
It is grant writing season (at least for me). My projects are essentially adapting Bayesian statistical methods in infectious disease epidemiology problems. But it seems to me that in order to make them competitive I need to add fancy words like learning and artificial intelligence...
#StatsSky
#StatsSky