Value of Forecasting in healthcare Planning

class: inverse
background-image:  linear-gradient(to right, rgba(50, 50, 50, .5), rgba(50, 50, 50, .4)),url("resources/titlepage.png")
background-size: cover

.title[Value of Forecasting in healthcare Planning]

.bottom[
### Bahman Rostami-Tabar (<svg style="height:0.8em;top:.04em;position:relative;fill:#1da1f2;" viewBox="0 0 512 512"><path d="M459.37 151.716c.325 4.548.325 9.097.325 13.645 0 138.72-105.583 298.558-298.558 298.558-59.452 0-114.68-17.219-161.137-47.106 8.447.974 16.568 1.299 25.34 1.299 49.055 0 94.213-16.568 130.274-44.832-46.132-.975-84.792-31.188-98.112-72.772 6.498.974 12.995 1.624 19.818 1.624 9.421 0 18.843-1.3 27.614-3.573-48.081-9.747-84.143-51.98-84.143-102.985v-1.299c13.969 7.797 30.214 12.67 47.431 13.319-28.264-18.843-46.781-51.005-46.781-87.391 0-19.492 5.197-37.36 14.294-52.954 51.655 63.675 129.3 105.258 216.365 109.807-1.624-7.797-2.599-15.918-2.599-24.04 0-57.828 46.782-104.934 104.934-104.934 30.213 0 57.502 12.67 76.67 33.137 23.715-4.548 46.456-13.32 66.599-25.34-7.798 24.366-24.366 44.833-46.132 57.827 21.117-2.273 41.584-8.122 60.426-16.243-14.292 20.791-32.161 39.308-52.628 54.253z"/></svg>[@Bahman_R_T](https://twitter.com/Bahman_R_T)) 
### Daniel Gartner (<svg style="height:0.8em;top:.04em;position:relative;fill:#1da1f2;" viewBox="0 0 512 512"><path d="M459.37 151.716c.325 4.548.325 9.097.325 13.645 0 138.72-105.583 298.558-298.558 298.558-59.452 0-114.68-17.219-161.137-47.106 8.447.974 16.568 1.299 25.34 1.299 49.055 0 94.213-16.568 130.274-44.832-46.132-.975-84.792-31.188-98.112-72.772 6.498.974 12.995 1.624 19.818 1.624 9.421 0 18.843-1.3 27.614-3.573-48.081-9.747-84.143-51.98-84.143-102.985v-1.299c13.969 7.797 30.214 12.67 47.431 13.319-28.264-18.843-46.781-51.005-46.781-87.391 0-19.492 5.197-37.36 14.294-52.954 51.655 63.675 129.3 105.258 216.365 109.807-1.624-7.797-2.599-15.918-2.599-24.04 0-57.828 46.782-104.934 104.934-104.934 30.213 0 57.502 12.67 76.67 33.137 23.715-4.548 46.456-13.32 66.599-25.34-7.798 24.366-24.366 44.833-46.132 57.827 21.117-2.273 41.584-8.122 60.426-16.243-14.292 20.791-32.161 39.308-52.628 54.253z"/></svg>[@drdanielgartner](https://twitter.com/drdanielgartner))

]

---

## Outline
<hr>

- How forecasts can be used to inform planning?

- Data & forecasting models

- Scheduling scheduling and utility measures

- Evaluate forecast accuracy

- .grey[What if we don't  bother with forecasting methods?]

---
## Outline
<hr>

- .remember[How forecast can be used to inform planning?]

- Data & forecasting models

- Scheduling scheduling and utility measures

- Evaluate forecast accuracy

- What would be the performance of planning if we plan based on forecasts?

- .grey[What if we don't  bother with forecasting methods?]

---
class: inverse, middle
background-image: linear-gradient(to right, rgba(50, 50, 50, .6), rgba(50, 50, 50, .6)), url("resources/as-staff.jpg")
background-size: cover

.title[How many nurses/doctors and When?]

- .title[Demand/capacity planning]
- .title[Staffing]
- .title[Scheduling]

---
## Forcasts can inform various planning horizons

Decision needs to be made to absorb demand uncertainty by providing effective and efficient staffing

- 6 months

- 42 days

- 48 hours

---
class: center middle

## Barriers in using forecasting

- Among other factors, .remember[disconnection] between forecasting and decision-making plays a role

- The vale/quality of a forecast should be determined by a decision it supports, something actionable

- Presenting both forecast accuracy and its impact can help to use forecasting to improve decision making

---
class: inverse center middle

.title[Forecasts are worthless until they lead to actions
]
---
## Aim of this project

- Evaluate the impact of forecasting on utility measure such as cost, overtime  staff, using emergency resources and waiting time

- Evaluate the use of probabilistic forecasting in planning

- What happens if you don't bother with forecasting methods

---
## Outline
<hr>

- How forecast can be used to inform planning?

- .remember[Data & forecasting models]

- Scheduling scheduling and utility measures

- Evaluate forecast accuracy

- What would be the performance of planning if we plan based on forecasts?

- .grey[What if we don't  bother with forecasting methods?]

---

---
## Data

- Data is hospital admissions/arrivals 
- 5 years of arrival time
- Public holidays, school holidays and rugby
- We use 365 days to test the performance of models
- We forecast for 48 hours ahead

---
<img src="p_hafhour.png" width="80%" style="display: block; margin: auto;" />

---

---

---
## Forecasting methods

- Poisson Regression
    - Autoregressive lags, daily, weekly and yearly seasonality, Trend 
    
- Prophet
    -  daily, weekly and yearly seasonality, Trend
    
- ARIMA

- Exponential Smoothing

- Seasonal Naive

---
## Outline
<hr>

- How forecast can be used to inform planning ?

- Data & forecasting models

- .remember[Shift scheduling and utility measures]

- Evaluate forecast accuracy using statistical measures

- What would be the performance of planning if we plan based on forecasts?

- What if we don't  bother with forecasting methods?

---

---

---

---

---
## Mathematical model

---
## Outline
<hr>

- How forecast can be used to inform planning?

- Data & forecasting models

- Shift scheduling and utility measures]

- .remember[Evaluate forecast accuracy using statistical measures]

- What would be the performance of planning if we plan based on forecasts?

- What if we don't  bother with forecasting methods?

---

---

---

---
<img src="p_nurse_distribution.png" width="80%" style="display: block; margin: auto;" />

---
<img src="utility_over_time.png" width="80%" style="display: block; margin: auto;" />

---
<img src="utility_over_time_distribution.png" width="80%" style="display: block; margin: auto;" />

---
<img src="cost_point_forecast.png" width="80%" style="display: block; margin: auto;" />

---
<img src="p_cost_distribution.png" width="80%" style="display: block; margin: auto;" />

---
.center[
.pull-left[
### Planning based on point forecast
<img src="utility_over_time_distribution.png" width="100%" style="display: block; margin: auto;" />
]

.pull-right[
### Planning based on 90% quantile
<img src="utility_over_time_distribution_90.png" width="100%" style="display: block; margin: auto;" />
]
]
---

.center[
.pull-left[
### Planning based on point forecast

<img src="p_cost_distribution.png" width="100%" style="display: block; margin: auto;" />
]

.pull-right[
### Planning based on 90% quantile
<img src="p_cost_distribution_90.png" width="100%" style="display: block; margin: auto;" />
]
]

---
## Next steps

- Include waiting time in the utility measure
- Evaluate the value for staffing (forecast for 42 days ahead)

---
class: inverse, top
background-image: linear-gradient(to right, rgba(300, 300, 300, .4), rgba(300, 300, 300, .5)), url("resources/questions.jpg")
background-size: cover

.hand-large[Thank you]

.remember[.larger[Question?]]