Example occupancy analysis#
Consider the same SSU example described in the Getting started with hillmaker tutorial. Now let’s do a little more involved analysis. Let’s assume that a renovation project is planned which will temporarily take 50 out of the 110 SSU beds out of service. The mitigation plan is to identify a subset of patient types served by the SSU that can be routed to an alternate holding area. We can use hillmaker to evaluate different subsets of patients to move with respect to the reduced capacity of the SSU during the renovation.
import itertools
import pandas as pd
import hillmaker as hm
ssu_stopdata = 'https://raw.githubusercontent.com/misken/hillmaker-examples/main/data/ssu_2024.csv'
# ssu_stopdata = './data/ssu_2024.csv'
stops_df = pd.read_csv(ssu_stopdata, parse_dates=['InRoomTS','OutRoomTS'])
stops_df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 59877 entries, 0 to 59876
Data columns (total 5 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 PatID 59877 non-null int64
1 InRoomTS 59877 non-null datetime64[ns]
2 OutRoomTS 59877 non-null datetime64[ns]
3 PatType 59877 non-null object
4 LOS_hours 59877 non-null float64
dtypes: datetime64[ns](2), float64(1), int64(1), object(1)
memory usage: 2.3+ MB
What are all the patient types?
patient_types = stops_df['PatType'].unique()
patient_types
array(['IVT', 'MYE', 'CAT', 'ART', 'OTH'], dtype=object)
Let’s start by running a scenario that includes all of the patient types.
ssu_all = hm.Scenario(scenario_name='ssu_all',
data=stops_df,
in_field='InRoomTS',
out_field='OutRoomTS',
start_analysis_dt='2024-01-02',
end_analysis_dt='2024-09-30',
cat_field='PatType',
bin_size_minutes=60,
cap=110,
verbosity=1,
)
ssu_all.make_hills()
2024-01-16 09:53:01,961 - hillmaker.hills - INFO - Starting scenario ssu_all
2024-01-16 09:53:10,900 - hillmaker.summarize - INFO - Created nonstationary summaries - ['PatType']
2024-01-16 09:53:12,037 - hillmaker.summarize - INFO - Created nonstationary summaries - []
2024-01-16 09:53:12,071 - hillmaker.summarize - INFO - Created stationary summaries - ['PatType']
2024-01-16 09:53:12,083 - hillmaker.summarize - INFO - Created stationary summaries - []
2024-01-16 09:53:13,416 - hillmaker.hills - INFO - bydatetime and summaries by datetime created (seconds): 11.4386
2024-01-16 09:53:14,222 - hillmaker.plotting - INFO - Full week plots created (seconds): 0.8039
2024-01-16 09:53:14,222 - hillmaker.hills - INFO - Total time (seconds): 12.2449
ssu_all.get_plot()
Ok, now we need to temporarily close 50 beds. Which patient populations can we move out of the SSU temporarily in order to deal with the reduced capacity? Of course we want to try to use the remaining capacity of 60 SSU beds as efficiently as possible.
One simple thing we could do is to run hillmaker with different combinations of patient types excluded from the analysis and then look at the resulting plots to see if a few scenarios stand out as reasonable alternatives. The cats_to_exclude parameters is useful for this type of analysis.
We can use the itertools library to generate combinations of patient types to consider rerouting. Each combination constitutes a different scenario. We will try combinations of up to three patient types. for example, we can generate a list of all combinations of two patient types like this:
[_ for _ in itertools.combinations(patient_types, 2)]
[('IVT', 'MYE'),
('IVT', 'CAT'),
('IVT', 'ART'),
('IVT', 'OTH'),
('MYE', 'CAT'),
('MYE', 'ART'),
('MYE', 'OTH'),
('CAT', 'ART'),
('CAT', 'OTH'),
('ART', 'OTH')]
At a high level, our analysis plan is:
Create a dictionary of hillmaker inputs that are fixed across all the scenarios.
Iterate over all the potential combinations of patients to be rerouted.
create a new hillmaker scenario based on the current iteration’s patient type combination
run hillmaker
create weekly occupancy plot for this scenario
Review all of the occupancy plots to eliminate obviously bad scenarios
Decide among the candidate scenarios based on the hillmaker results as well as other non-capacity related consideratoins.
Step 1 - Fixed inputs#
# Create dict of fixed inputs
ssu_inputs = {'data': stops_df,
'in_field': 'InRoomTS',
'out_field': 'OutRoomTS',
'start_analysis_dt': '2024-01-02',
'end_analysis_dt': '2024-03-30',
'cat_field': 'PatType',
'bin_size_minutes': 60,
'cap': 60,
'verbosity': 0,
}
Step 2 - Process each scenario with hillmaker#
# Generate the list of combinations of patient types to exclude
patient_types = stops_df['PatType'].unique()
exclusions = [[_ for _ in itertools.combinations(patient_types, i)] for i in range(1,4)]
# Create container dictionary to store the weekly plots
plots = {}
# Main loop over number of patient types to exclude
for num_cats_excluded in range(1, 4):
# Loop over each combination of patient types excluded (the -1 is for the zero index)
for exclusion in exclusions[num_cats_excluded - 1]:
cats_to_exclude = exclusion
print(f'cats_to_exclude = {cats_to_exclude}')
scenario_name = f'no_{"_".join(list(exclusion))}'
print(f'scenario_name = {scenario_name}')
# Create scenario object
scenario = hm.create_scenario(params_dict=ssu_inputs,
scenario_name=scenario_name, cats_to_exclude=cats_to_exclude)
# Make the hills
scenario.make_hills()
# Create the weekly plot (they're also being exported)
plots[scenario_name] = scenario.make_weekly_plot(ylabel='Patients', plot_export_path='./output')
cats_to_exclude = ('IVT',)
scenario_name = no_IVT
cats_to_exclude = ('MYE',)
scenario_name = no_MYE
cats_to_exclude = ('CAT',)
scenario_name = no_CAT
cats_to_exclude = ('ART',)
scenario_name = no_ART
cats_to_exclude = ('OTH',)
scenario_name = no_OTH
cats_to_exclude = ('IVT', 'MYE')
scenario_name = no_IVT_MYE
cats_to_exclude = ('IVT', 'CAT')
scenario_name = no_IVT_CAT
cats_to_exclude = ('IVT', 'ART')
scenario_name = no_IVT_ART
cats_to_exclude = ('IVT', 'OTH')
scenario_name = no_IVT_OTH
cats_to_exclude = ('MYE', 'CAT')
scenario_name = no_MYE_CAT
cats_to_exclude = ('MYE', 'ART')
scenario_name = no_MYE_ART
cats_to_exclude = ('MYE', 'OTH')
scenario_name = no_MYE_OTH
cats_to_exclude = ('CAT', 'ART')
scenario_name = no_CAT_ART
cats_to_exclude = ('CAT', 'OTH')
scenario_name = no_CAT_OTH
cats_to_exclude = ('ART', 'OTH')
scenario_name = no_ART_OTH
cats_to_exclude = ('IVT', 'MYE', 'CAT')
scenario_name = no_IVT_MYE_CAT
cats_to_exclude = ('IVT', 'MYE', 'ART')
scenario_name = no_IVT_MYE_ART
cats_to_exclude = ('IVT', 'MYE', 'OTH')
scenario_name = no_IVT_MYE_OTH
cats_to_exclude = ('IVT', 'CAT', 'ART')
scenario_name = no_IVT_CAT_ART
cats_to_exclude = ('IVT', 'CAT', 'OTH')
scenario_name = no_IVT_CAT_OTH
cats_to_exclude = ('IVT', 'ART', 'OTH')
scenario_name = no_IVT_ART_OTH
cats_to_exclude = ('MYE', 'CAT', 'ART')
scenario_name = no_MYE_CAT_ART
cats_to_exclude = ('MYE', 'CAT', 'OTH')
scenario_name = no_MYE_CAT_OTH
cats_to_exclude = ('MYE', 'ART', 'OTH')
scenario_name = no_MYE_ART_OTH
cats_to_exclude = ('CAT', 'ART', 'OTH')
scenario_name = no_CAT_ART_OTH
Step 3 - Review weekly plots for promising scenarios#
By reviewing the weekly occupancy plots, we can identify a handful of promising scenarios.
For example, excluding just the CAT patients clearly doesn’t free up enough capacity.
plots['no_CAT']
Conversely, excluding IVT, ART, and OTH, results in too much unused capacity.
plots['no_IVT_ART_OTH']
The two most promising scenarios would seem to be:
CAT, ART, OTH
IVT, MYE
plots['no_CAT_ART_OTH']
plots['no_IVT_MYE']
Of course, this is just a starting point for actually making a decision. There are other factors that will influence the feasibility of these (or other) scenarios. For example, a scenario involving moving only the IVT patients leads to this projected occupancy. Perhaps with some changes in scheduling such as moving some volume from Mondays and Fridays to mid-week, this scenario might be workable.
plots['no_IVT']
