# Working log {.unnumbered}

Here is the Future work for TSA and Randy's thesis

## 20231101

1. Including other features and outcomes for matching

2. Using both pain for informative outcome for outcome inference

2\*. pain and dash two outcomes

3. ✅ 365 days cut off this is the last step of data cleaning

4. ✅ Insurance company, and the worker report worse

5. ✅ Cleaning multiple surgeries

- filter out the patient with only one

- filter out the two situations

- gap between the two surgeries are short or with few outcomes

- two adjacent surgeries without outcomes in between

- the second surgery without any further outcomes after

## 20231106

6. ✅ Cleaning the pain dataset

7. ✅ Table or histogram for the outcome number, timemax, and pain

## 20231109

8. How to use the tsa data.

- filter the patients with the first outcome in a week after surgery

- use imputation inla-rw or inla-bspline as the baseline

- ask Stef whether brokenstick or multiple-imputation is powerful enough to provide

- test on the patients with first outcome in a week (pick out the patients)

9. ✅ The pain and outcome at the same time ??? - most with both outcomes, very few no at the same time

10. only keep the first surgery for every one

11. the baseline will be `what ever the first outcome value is` and

include the time gap between surgery and the first outcome