class: center, middle, inverse, title-slide # COVID-19 modelling in Ontario:<br>past, present, future ### <br>Irena Papst, PhD<br>Postdoctoral Fellow<br>Department of Mathematics & Statistics<br>McMaster University --- class: middle, center Slides available at http://papsti.github.io/talks/2022-03-11_TDW.html --- class: middle, center I wish to acknowledge this land on which the University of Toronto operates. For thousands of years it has been the traditional land of the **Huron-Wendat**, the **Seneca**, and the **Mississaugas of the Credit.** Today, this meeting place is still the home to many Indigenous people from across Turtle Island and I am grateful to have the opportunity to present today on this land. ??? - Indigenous land acknowledgment statement adapted from [Indigenous U of T](https://indigenous.utoronto.ca/about/land-acknowledgement/) --- class: center, middle, inverse ## Forecasting is an important tool<br>in controlling the spread of COVID-19 ??? - short-term forecasting of COVID-19 has proven itself to be an **enormously useful** tool for **public health planning** throughout this pandemic --- class: center, middle <img src="data:image/png;base64,#figs/collages/collages.001.png" width="900" style="display: block; margin: auto;" /> ??? - we've seen it time and time again - fun fact: some of those curves come from my research group, which regularly submit forecasts to --- class: middle, center <img src="data:image/png;base64,#figs/osat.png" width="1300" style="display: block; margin: auto;" /> ??? the ontario science advisory table, where the forecasts you'll see today have been submitted to, to both brief the gov and the public equally --- class: center, middle, inverse ## Forecasting is difficult,<br>but (usually) doable ??? - lots of moving parts, the situation is shifting on the ground, but if you're keeping on top of it, you can make useful forecasts on the fly - let's look at an example now --- class: inverse, center, middle ## 🔮 Forecast from 21 Feb 2021 🔮 ??? - this is for the province of ontario, canada - the first time variants became a factor in ontario was in early 2021 - not sure about you all, but this feels like years ago to me, so let's first remind ourselves what was going on at the time --- <img src="data:image/png;base64,#figs/feb-context-1.png" width="1344" style="display: block; margin: auto;" /> ??? - in the **lead up** to the forecast date, marked here with a **dashed line**, we saw a slow and steady increase in **infection reports** - and while **2000/3000 infections** reported in a province of almost **15 million** may not seem like much, the **strain** on the healthcare system was beginning to show --- <img src="data:image/png;base64,#figs/feb-context-2.png" width="1344" style="display: block; margin: auto;" /> ??? ----- - you can see **ICU occupancy** had also been rising with infection reports --- <img src="data:image/png;base64,#figs/feb-context-3.png" width="1344" style="display: block; margin: auto;" /> ??? - and we had actually surpassed the **pre-COVID ICU capacity** threshold for the entire province - while we did expand ICU capacity in response to the pandemic, it was achieved through **hugely disruptive measures**, including **cancellations of "elective" surgeries** that sound optional, but may actually be lifesaving down the line - this was the beginning of a **potential crisis** in our hospitals, and we had to act --- <img src="data:image/png;base64,#figs/feb-context-4.png" width="1344" style="display: block; margin: auto;" /> ??? - in response to rising infection reports and ICU occupancy, the government implemented **several interventions**, namely - lockdowns in Toronto and Peel in late november, the **two most populous health regions in the province** (representing about 30% of the provincial population) - followed by a **province-wide shutdown** on 26 december - and then ultimately a **province-wide stay-at-home order** in mid-January --- <img src="data:image/png;base64,#figs/feb-context-5.png" width="1344" style="display: block; margin: auto;" /> ??? - the combination of these interventions finally began **decreasing** infection reports and **pretty sharply** compared to the original incline, though **decline in ICU occupancy** was **slower** --- <img src="data:image/png;base64,#figs/feb-context-6.png" width="1344" style="display: block; margin: auto;" /> ??? - the stay-at-home order was **set to expire** in early march, and since things were looking good, it was looking **likely** that the province **wouldn't extend it** - so the **question** at the time (as it usually is in these forecasts) was --- class: inverse, middle, center # Is it safe to<br>loosen restrictions? --- <img src="data:image/png;base64,#figs/feb-context-6.png" width="1344" style="display: block; margin: auto;" /> ??? - **just** looking at the trends in infection reports and ICU occupancy, you may think *"yes, it should be safe to reopen!"* - ICU occupancy was **still high** but **trending downward**, and we know the ICU trend is **lagged** from infection reports---people enter the ICU some time after testing positive and then they may stay in the ICU for a while - but there's **another piece of the puzzle** that isn't clear from infection reports or ICU occupancy --- class: center <img src="data:image/png;base64,#figs/feb-context-7.png" width="1344" style="display: block; margin: auto;" /> ??? - **under the hood**, the Alpha varaint (aka strain **B.1.1.7**) had made its way into the province - this variant is about **50% more transmissible** compared to resident strains at the time - it was first confirmed in the province **coincidentally** on the same day as the province-wide shutdown, **26 december** - by the forecast date, it represented about **20%** of infection reports - we were particularly concerned about Alpha because **despite** the fact that our public health interventions were **successfully driving infection reports down pretty rapidly**, Alpha was finding a way to **increase its frequency** over this period... _scary!_ - we also saw what alpha had done in the UK, which was about a month ahead of us in this surge, and so we were quite concerned -- Trends in <span style="font-weight:600;">infection reports</span> and <span style="color:#6F3B42;font-weight:600;">ICU occupancy</span> favoured reopening,<br>while <span style="color:#CE9A70;font-weight:600;">Alpha</span> threatened it. --- class: center <br> <br> <br> <br> <img src="data:image/png;base64,#figs/feb-forecast-1.png" width="1344" style="display: block; margin: auto;" /> ??? - so we fit transmission to the infection report data (the **black curve**) and then - so we made **two forecasts** at the time: 1. one where we **explicitly modelled the increasing frequency** of the Alpha variant through both the fit and the forecast period (**extended model**) 2. another where we detected **whatever effect** the variant was having on infection reports by fitting the transmission rate as usual, then we just **propagated that effect forward**, with no change to Alpha frequency going forward (**base model**) - the curve is the median forecast in ecah case and the band around each curve represents 95% confidence intervals - and here is **what happened** actually happened --- class: center <br> <br> <br> <br> <img src="data:image/png;base64,#figs/feb-forecast-2.png" width="1344" style="display: block; margin: auto;" /> ??? - these **hollow points** are still infection reports, just those observed after the forecast date - you can see that the forecast using the **extended model**, with the effect of the variant, **caught the third wave takeoff** - while the forecast using the base model showed some take-off, but **grossly underestimated the magnitude of the uptick** resulting from the measures relaxing -- <span style="font-size: 30px;display:block; margin-top:85px;">Infection reports *alone* could not signal<br>the start of the <span style="color:#CE9A70;font-weight:600;">third wave</span>.</span> ??? - infection reports alone... - and that **including the variant** in our model was **crucial** for this forecast - now this is an example of a successful forecast, and in order to help you appreciate how much more difficult forecasting has gotten recently, --- class: center, middle, inverse # How do we make these forecasts? ??? - let's take a look under the hood at how the sausage is made... how exactly do we make these forecasts? --- class: center, middle <img src="data:image/png;base64,#figs/collages/collages.002.png" width="1000" style="display: block; margin: auto;" /> --- class: center, middle <img src="data:image/png;base64,#figs/collages/collages.003.png" width="1000" style="display: block; margin: auto;" /> ??? - so let's start with what the data is we're fitting to --- ## 📊 Data <img src="data:image/png;base64,#figs/infection-reports.png" height="300" style="display: block; margin: auto;" /> Number of (reported) **cases** or **infection reports** in Ontario ??? this is usually what we are trying to forecast for a few weeks into the future -- - **underestimate** of actual infections, **delay** of 1-2 weeks - depend on **testing availability** and **behaviour** -- Can also fit to and forecast **hospitalizations**, **ICU admissions**, **deaths** simultaneously ??? - the other input into the fitting step is --- <img src="data:image/png;base64,#figs/collages/collages.004.png" width="1000" style="display: block; margin: auto;" /> --- ## 💭 Model We use a **compartmental** epidemiological model (SEIR framework) ??? - this jargon may or may not mean something to you, so let's break it down and make sure we're all on the same page -- <img src="data:image/png;base64,#figs/flowchart1.png" height="370" style="display: block; margin: auto;" /> ??? - SEIR refers to the main compartments in this model: susceptible, exposed, infected, recovered - the population is **partitioned** into one of these states - **everyone starts in the susceptible class** and then individuals can get **exposed** to (and therefore infected by) SARS-CoV-2 - and then there are downstream outcomes ----- - each of these arrows represents a **flow rate** specified with various parameters -- New infections (and thus **infection reports**, our forecast target) are governed by **transmission**: susceptible `\(\rightarrow\)` exposed ??? however, new infections (and thus infection reports, the data we're usually calibrating to) are governed by transmission, so the susceptible to exposed pathway is the one we focus on --- ## 🦠 Transmission ??? so what we generally care about getting right is transmission in the population (key to our model fitting) -- We model **transmission** as $$ \quad S(t) \xrightarrow{(\color{#1E3765}{\text{force of infection}}) \times S(t)} E(t) $$ where `\(S(t) =\)` susceptibles at time t, and `\(E(t) =\)` exposeds. ??? ----- - transmission converts susceptibles to exposeds - the **per capita rate** with which this **conversion** happens is called the **force of infection** -- Then we assume $$ \color{#1E3765}{\text{force of infection} = \beta \times \frac{I(t)}{N}} $$ for `\(I(t) =\)` infecteds and `\(N\)`, the total population size. ??? ----- - the **basic structure** of the force of infection in our model (though it's actually more complicated because we have different types of infecteds) -- - `\(\color{#1E3765}{\beta}\)`: transmission rate (# of **transmission opportunities** per individual per day, depends on the average **contact rate** between individuals) ??? ----- - **transmission opportunities** -- - `\(\color{#1E3765}{\frac{I(t)}{N}}\)`: probability of **encountering an infective** during a transmission opportunity at time `\(\color{#1E3765}{t}\)`, (assuming **homogeneous mixing**) ??? ----- - probability of encountering an infective (**encounters are random**) - the key parameter we want to fit is `\(\beta\)` --- <img src="data:image/png;base64,#figs/collages/collages.005.png" width="1000" style="display: block; margin: auto;" /> --- ## 🔧 Calibration Transmission, `\(\color{#1E3765}{\beta}\)`, does **not** remain constant over time! ??? - does anyone want to take a guess at what might change transmission over time? -- Depends on: -- - public health **measures** -- - population **behaviour** -- - mix of **variants** floating around -- - **vaccine** coverage -- <br> Assume `\(\color{#1E3765}{\beta(t)}\)` is **piecewise-constant**, and re-estimate when there has been a **significant change in public health measures and/or behaviour** (e.g. restrictions loosening, stay-at-home order). -- (Effects of **variants** and **vaccines** are **incorporated explicitly**—we're not trying to fit this.) --- ## 🔧 Calibration <img src="data:image/png;base64,#figs/betademo-1.png" width="1344" style="display: block; margin: auto;" /> ??? ----- - here is an example of what a fitted beta(t) might look like --- <img src="data:image/png;base64,#figs/collages/collages.006.png" width="1000" style="display: block; margin: auto;" /> --- ## 🔮 Forecasting <img src="data:image/png;base64,#figs/betademo-1.png" width="1344" style="display: block; margin: auto;" /> We make **assumptions** on `\(\color{#1E3765}{\beta(t)}\)` in the forecast period. ??? - we make... --- ## 🔮 Forecasting <img src="data:image/png;base64,#figs/betademo2-1.png" width="1344" style="display: block; margin: auto;" /> We make **assumptions** on `\(\color{#1E3765}{\beta(t)}\)` in the forecast period. Start with **latest** `\(\color{#1E3765}{\beta}\)` (from last estimation window). ??? - start with... --- ## 🔮 Forecasting <img src="data:image/png;base64,#figs/betademo-1.png" width="1344" style="display: block; margin: auto;" /> We make **assumptions** on `\(\color{#1E3765}{\beta(t)}\)` in the forecast period. Start with **latest** `\(\color{#1E3765}{\beta}\)` (from last estimation window). In **forecast period**, either: ??? - in forecast period... --- ## 🔮 Forecasting <img src="data:image/png;base64,#figs/betademo3-1.png" width="1344" style="display: block; margin: auto;" /> We make **assumptions** on `\(\color{#1E3765}{\beta(t)}\)` in the forecast period. Start with **latest** `\(\color{#1E3765}{\beta}\)` (from last estimation window). In **forecast period**, either: 1. use latest `\(\color{#1E3765}{\beta}\)` (**status quo** forecast) ??? - use latest... --- ## 🔮 Forecasting <img src="data:image/png;base64,#figs/betademo4-1.png" width="1344" style="display: block; margin: auto;" /> We make **assumptions** on `\(\color{#1E3765}{\beta(t)}\)` in the forecast period. Start with **latest** `\(\color{#1E3765}{\beta}\)` (from last estimation window). In **forecast period**, either: 1. use latest `\(\color{#1E3765}{\beta}\)` (**status quo** forecast) 2. increase (or decrease) transmission by **x%** relative to latest `\(\color{#1E3765}{\beta}\)` ??? - increase (or decrease)... - that's the general procedure we employ - and this all worked very well for a while, as we saw from that first forecast we looked at... but then --- class: inverse, center, middle # 🦠 Omicron 🦠 ??? - has been notoriously difficult to forecast for a few reasons - a quick recap --- class: center, middle <img src="data:image/png;base64,#figs/omi-timeline-1.png" width="1000" style="display: block; margin: auto;" /> --- class: center, middle <img src="data:image/png;base64,#figs/omi-timeline-2.png" width="1000" style="display: block; margin: auto;" /> --- class: inverse, center, middle ## How safe is the 31 Jan reopening? ??? - in january, the question we were focused on was --- ## ⛰ Recent forecasting challenges ??? - here are some of the challenges we were facing -- **Synchronized surges** of Omicron across similar jurisdictions around the world <img src="data:image/png;base64,#figs/surge-sync.png" width="700" style="display: block; margin: auto;" /> ??? -didn't really have the benefit of analyses and parameter estimates from elsewhere for this new variant - couldn't compare to south africa - (in the alpha wave, we benefited greatly from parameter estimates from the UK, which was about a month ahead of us in their surge) --- ## ⛰ Recent forecasting challenges Disease surveillence has been **significantly limited** -- **Tests completed** <img src="data:image/png;base64,#figs/test-vol.png" width="700" style="display: block; margin: auto;" /> ??? - relatively small number given we were taking off exponentially... low supply or low demand? -- **Test positivity** <img src="data:image/png;base64,#figs/percent-pos.png" width="700" style="display: block; margin: auto;" /> ??? - low supply - so much demand, the gov changed criteria for who in the general population can get a PCR test... need to be high risk of severe infection now (even more underreporting and bias towards detecting severe infection) --- ## ⛰ Recent forecasting challenges <br> Calibrating to **acute care** (non-ICU hospital) **occupancy alone** after 15 Dec -- <br> Needed expected **severity** (% of infections that are hospitalized) to calibrate transmission from _hospitalizations_ ??? this was a big question at the time, and there was a lot of debate as to how mild this thing would be in Ontario's population -- <span style="font-size: 30px;display:block; margin-top:120px;"> **Identifiability issue**: at least two distinct cases fit the hospitalization data equally well </span> --- ## 🔮 "Forecast" from 18 Jan 2022 -- <img src="data:image/png;base64,#figs/forecast_status_quo-1.png" width="700" style="display: block; margin: auto;" /> ??? - just a status quo forecast --- ## 🔮 "Forecast" from 18 Jan 2022 <img src="data:image/png;base64,#figs/forecast_status_quo2-1.png" width="700" style="display: block; margin: auto;" /> ??? - the difference between these two worlds is how many people actually got infected in the omicron wave, a.k.a. how much natural immunity did we build up in the population - if a lot of people actually got sick in that wave, reopening would pose less of a risk to hospital capacity - if not many people got sick, we could have been in for a bigger hospiatlization surge -- At the time, we didn't know **which world we were in**,<br>so we don't know **how safe it was to reopen** ??? - but then we got some reliable severity estimates from PHO, which we fed into the model so that we could make a real forecast --- ## 🔮 Forecast from 26 Jan 2022 -- <img src="data:image/png;base64,#figs/forecast_2022-01-26-1.png" width="800" style="display: block; margin: auto;" /> ??? - as you can see, there were a lot of assumptions that went into this one... didn't know the effect of schools nor the effect of 31 jan reopening, so we made scenarios - what happened? --- ## 🔮 Forecast from 26 Jan 2022 <img src="data:image/png;base64,#figs/forecast_2022-01-26-2.png" width="800" style="display: block; margin: auto;" /> ??? - does that mean that we built up substantial natural immunity in the population? maybe. can't say that's what's causing thise downturn. could be a lot of factors, including population behaviour, those being highest risk of spreading being recently immune, booster protection is better than we assumed in the model, etc. - may simply be that the relaxation we did on 31 jan wasn't that risky for hospitalizations! --- ## 👀 Looking forward -- - **goals are changing** 🎯 controlling ~~spread~~ hospital occupancy ??? - this mentality leaves a number of people behind: not everyone can afford to take their chances with COVID and hope it's mild - even mild cases can lead to serious outcomes (which also continue to tax the healthcare system + peoples' quality of life): long COVID, premature brain ageing -- - "we're **done**" & "this is the **best we can do**" 😩 ??? - critical mass seems to have shifted towards being done with the pandemic... but the pandemic is not done with us - i understand this mentality... lots of people now know someone who had mild COVID, feels like much less of a threat (and it is, thanks to vaccination!), so we're doing things like -- - taking more risks 🤞 - lifting restrictions (reopenings) 🎉 - reduced testing requirements 🔓 ??? - this is not all bad, we do have to live our lives, but we need to be smart about it -- - 🚨 **this disease will not just disappear** 🚨 ??? - smallpox is the only disease we've successfully eradicated in humans... -- - did H1N1, HIV, avian flu disappear? ??? - all these other pandemic pathogens you may have heard of are all still floating around! - but that doesn't mean they're necessarily a huge threat to us anymore - SARS-CoV-2 is now part of our ecosystem, which is not to say that we may not one day eradicate it, we're just not there yet - need to be ready and willing to act when the situation worsens, and relax when it gets better - if we barrel into normalcy by force, without a solid strategy, we will continue to bear a needlessly high disease burden - we're not soley at the mercy of this virus and i want to leave you with a bit of optimism here, because i really do think/hope that the worst is over --- ## 👀 Looking forward - **plan** for occasional COVID outbreaks 📝 ??? - COVID flare ups will happen, we need to have a plan for them, to keep the most vulnerable in our society safe (just like with influenza) -- - temporary **masking** 😷 and **testing** 🧪 advisories -- - distribution of free **masks** and **rapid tests** -- - normalize easy **protective behaviours** ⚠️ when the situation worsens -- - masking in public spaces - rapid testing before particularly large gatherings ??? - be prepared to mask - be prepared to test - getting good masks and rapid tests into peoples' medicine cabinets before outbreaks happen - need to normalize easy protective behaviours - masking in public spaces during COVID season as a normal thing to do (in east asian people routinely mask in public during flu season) - rapid testing before particularly large gatheringsthat we've seen can be superspreader events (e.g., indoor weddings, birthday parties, etc.) - but on order to do any of this, we need to have eyes on the situation -- - don't give up on **surveillence** 🔦 🔍 👀 -- - **wastewater** -- - ✅ changes in spread 📈 📉 ??? - great for tracking changes in trends (increase/decrease in spread) -- - ❌ number of infections ??? - difficult to infer number of infections (and therefore project hospitalizations) with much confidence - viral shedding into wastewater varies between individuals and may depend on severity of illness, vax status, variant -- - routine **workplace** testing 🏥 🧸 ??? - esp in higher risk settings (long-term care homes, daycares) - great use of rapid tests -- - **sentinel** surveillence 💂 ??? - enroll randomized households in free routine testing (randomize geographically, socioeconomically, racially), and bring the sampling to the people (e.g. REACT study UK, 150k participants each month) - and the message i'd like to leave you with is the following --- class: inverse, center, middle # 🔑 We need to learn to<br>live _with_ COVID,<br>not live _despite_ it 🔑 ??? - its tempting to do the latter, what i think a number of people are doing now, but if we want to keep our disease burdens down, we have to be a bit more strategic --- class: center, middle ## 🤝 Collaborators <img src="data:image/png;base64,#figs/collaborators.png" width="800" style="display: block; margin: auto;" /> ??? - with that, i'd like to thank my collaborators - david ben and jonathan are faculty at mcmaster, evan is also a postdoc - and in the bottom row we have mike li, now at PHAC, truly central to our forecasting effort, and steven walker who is a data scientist at canmod, a newly-established canadian mathematical modelling network --- class: center, middle <img src="data:image/png;base64,#figs/macpan.png" width="800" style="display: block; margin: auto;" /> [https://github.com/mac-theobio/McMasterPandemic](https://github.com/mac-theobio/McMasterPandemic) ??? - i'll also mention that the model and forecasting software are implemented as a **publicly-available R package**, which you can find on **Github** --- class: center <br> <br> <br> <br> <br> # 🔮 Forecast updates [https://mac-theobio.github.io/covid-19/](https://mac-theobio.github.io/covid-19/) <br> <br> <br> # Thank you for listening! 🙏 ??? - blog posts detailing our latest forecasts, as well as some i've discussed in this presentation (posts go into more detail than i've done here, if you're interested)