Exp test patterns

crcsim/experiment/README.md

@@ -55,7 +55,7 @@ The subdirectories and files in `scenarios/` must be uploaded to AWS S3 for the
 
 To upload the files to S3, run
 ```
-aws s3 cp ./scenarios s3://crcsim-exp-crccp-sensitivity01/scenarios --recursive
+aws s3 cp ./scenarios s3://crcsim-exp-test-patterns/scenarios --recursive
 ```
 *(Another note: this manual step is necessary because `boto3` does not include functionality to upload a directory to S3 recursively. Future experiments could improve this workflow by writing a function to upload the directory recursively in `prepare.py`. Or submit a patch to resolve https://github.com/boto/boto3/issues/358)*
 

crcsim/experiment/parameters.json

@@ -82,8 +82,8 @@
       "cost": 22,
       "proportion_perforation": 0,
       "cost_perforation": 0,
-      "compliance_rate_given_prev_compliant":     [ 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 ],
-      "compliance_rate_given_not_prev_compliant": [ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ]
+      "compliance_rate_given_prev_compliant":     [ 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 ],
+      "compliance_rate_given_not_prev_compliant": [ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ]
     },
     "Colonoscopy": {
       "proportion": 0.0,
@@ -98,8 +98,8 @@
       "cost": 700,
       "proportion_perforation": 0.001,
       "cost_perforation": 6487,
-      "compliance_rate_given_prev_compliant":     [ 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 ],
-      "compliance_rate_given_not_prev_compliant": [ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ]
+      "compliance_rate_given_prev_compliant":     [ 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 ],
+      "compliance_rate_given_not_prev_compliant": [ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ]
     }
   },
   "routine_tests": [ "FIT", "Colonoscopy" ],

crcsim/experiment/prepare.py

@@ -1,8 +1,8 @@
 import json
 import random
+from enum import Enum, unique
 from pathlib import Path
 from typing import Callable, Dict, List, Optional
-from copy import deepcopy
 
 import fire
 
@@ -20,6 +20,18 @@ def transform(self, transformer: Callable) -> "Scenario":
         return self
 
 
+@unique
+class ConditionalComplianceParam(Enum):
+    PREV_COMPLIANT = "compliance_rate_given_prev_compliant"
+    NOT_PREV_COMPLIANT = "compliance_rate_given_not_prev_compliant"
+
+
+@unique
+class Test(Enum):
+    FIT = "FIT"
+    COLONOSCOPY = "Colonoscopy"
+
+
 def prepare_experiment_dir(
     scenarios: List[Scenario],
     num_iterations: int,
@@ -87,148 +99,130 @@ def transform(params):
     return transform
 
 
-def transform_treatment_cost(stage: str, phase: str, value: int) -> Callable:
-    def transform(params):
-        params[f"cost_treatment_stage{stage}_{phase}"] = value
-
-    return transform
-
+def transform_conditional_compliance_rates(
+    test: Test, param: ConditionalComplianceParam, rates: List[float]
+) -> Callable:
+    """
+    Replaces a conditional compliance parameter with a new set of rates.
+    """
 
-def transform_repeat_compliance(rate: float, test: str) -> Callable:
     def transform(params):
-        params["tests"][test]["compliance_rate_given_prev_compliant"] = [
-            rate for _ in params["tests"][test]["compliance_rate_given_prev_compliant"]
-        ]
+        params["tests"][test.value][param.value] = rates
 
     return transform
 
 
-def transform_diagnostic_compliance(rate) -> Callable:
-    def transform(params):
-        params["diagnostic_compliance_rate"] = rate
-
-    return transform
-
+def transform_delayed_onset(test: Test, onset_age: int) -> Callable:
+    """
+    Transforms the necessary compliance parameters such that, for the given test,
+    compliance is 0.0 for all years before the given age, and 1.0 for all years
+    after the given age.
+    """
 
-def transform_surveillance_frequency(stage: str, frequency: int) -> Callable:
     def transform(params):
-        params[f"surveillance_freq_{stage}"] = frequency
+        start_age = params["tests"][test.value]["routine_start"]
+        # Testing years are inclusive, so add 1
+        testing_years = params["tests"][test.value]["routine_end"] - start_age + 1
+
+        # Initial compliance is 0
+        params["initial_compliance_rate"] = 0.0
+
+        # Then everyone remains noncompliant until the given age.
+        compliance_rate_given_not_prev_compliant = [0.0] * testing_years
+        onset_year = onset_age - start_age
+        compliance_rate_given_not_prev_compliant[onset_year] = 1.0
+        params["tests"][test.value][
+            "compliance_rate_given_not_prev_compliant"
+        ] = compliance_rate_given_not_prev_compliant
+
+        # Then everyone remains compliant after the given age.
+        # We can just set the compliance rate to 1.0 for all years, including those before
+        # the onset age, because initial compliance is 0.0.
+        compliance_rate_given_prev_compliant = [1.0] * testing_years
+        params["tests"][test.value][
+            "compliance_rate_given_prev_compliant"
+        ] = compliance_rate_given_prev_compliant
 
     return transform
 
 
-def transform_surveillance_end_age(age: int) -> Callable:
+def transform_routine_freq(test: Test, freq: int) -> Callable:
     def transform(params):
-        params["surveillance_end_age"] = age
+        params["tests"][test.value]["routine_freq"] = freq
 
     return transform
 
 
 def create_scenarios() -> List:
-    # For each health center, define the initial compliance rate in the baseline
-    # scenario and the implementation scenario.
-    initial_compliance = {
-        "fqhc1": (0.522, 0.593),
-        "fqhc2": (0.154, 0.421),
-        "fqhc3": (0.519, 0.615),
-        "fqhc4": (0.278, 0.374),
-        "fqhc5": (0.383, 0.572),
-        "fqhc6": (0.211, 0.392),
-        "fqhc7": (0.257, 0.354),
-        "fqhc8": (0.190, 0.390),
-    }
-    low_initial_stage_3_treatment_cost = 67_300
-    low_initial_stage_4_treatment_cost = 97_931
-    low_diagnostic_compliance_rate = 0.525
-    lower_repeat_compliance = 0.8
-    low_surveillance_freq_mild = 10
-    low_surveillance_freq_severe = 2
-    low_surveillance_end_age = 80
-    scenarios = []
-
-    for fqhc, rates in initial_compliance.items():
-        baseline = Scenario(
-            name=f"{fqhc}_baseline", params=get_default_params()
-        ).transform(transform_initial_compliance(rates[0]))
-        scenarios.append(baseline)
-
-        implementation = Scenario(
-            name=f"{fqhc}_implementation", params=get_default_params()
-        ).transform(transform_initial_compliance(rates[1]))
-        scenarios.append(implementation)
-
-        # Sensitivity Analysis 1.  Lower repeat compliance (note that the baseline runs stay the same)
-
-        test_name = "FIT"
-        implementation_lower_repeat_compliance = deepcopy(implementation)
-        implementation_lower_repeat_compliance.transform(
-            transform_repeat_compliance(lower_repeat_compliance, test_name)
-        )
-        implementation_lower_repeat_compliance.name = (
-            f"{fqhc}_implementation_lower_repeat_compliance"
-        )
-        scenarios.append(implementation_lower_repeat_compliance)
-
-        # Sensitivity analysis 2. Lower cost for stage III and stage IV initial phase
-        baseline_low_cost = deepcopy(baseline)
-        baseline_low_cost.transform(
-            transform_treatment_cost("3", "initial", low_initial_stage_3_treatment_cost)
-        ).transform(
-            transform_treatment_cost("4", "initial", low_initial_stage_4_treatment_cost)
-        )
-        baseline_low_cost.name = f"{fqhc}_baseline_low_initial_treat_cost"
-        scenarios.append(baseline_low_cost)
-
-        implementation_low_cost = deepcopy(implementation)
-        implementation_low_cost.transform(
-            transform_treatment_cost("3", "initial", low_initial_stage_3_treatment_cost)
-        ).transform(
-            transform_treatment_cost("4", "initial", low_initial_stage_4_treatment_cost)
-        )
-        implementation_low_cost.name = f"{fqhc}_implementation_low_initial_treat_cost"
-        scenarios.append(implementation_low_cost)
-
-        # Sensitivity analysis 3. Lower compliance with diagnostic colonoscopy
-        baseline_lower_compliance = deepcopy(baseline)
-        baseline_lower_compliance.transform(
-            transform_diagnostic_compliance(low_diagnostic_compliance_rate)
-        )
-        baseline_lower_compliance.name = f"{fqhc}_baseline_lower_diagnostic_compliance"
-        scenarios.append(baseline_lower_compliance)
-
-        implementation_lower_compliance = deepcopy(implementation)
-        implementation_lower_compliance.transform(
-            transform_diagnostic_compliance(low_diagnostic_compliance_rate)
-        )
-        implementation_lower_compliance.name = (
-            f"{fqhc}_implementation_lower_diagnostic_compliance"
-        )
-        scenarios.append(implementation_lower_compliance)
 
-        # TODO: Sensitivity analysis 4. Lower surveillance frequency and end age.
+    scenarios = []
 
-        baseline_lower_surveillance = deepcopy(baseline)
-        baseline_lower_surveillance.transform(
-            transform_surveillance_frequency("polyp_mild", low_surveillance_freq_mild)
-        ).transform(
-            transform_surveillance_frequency("polyp_severe", low_surveillance_freq_severe)
-        ).transform(
-            transform_surveillance_end_age(low_surveillance_end_age)
+    # Scenarios: 100% FIT compliance
+    always_compliant = Scenario(
+        name="always_compliant", params=get_default_params()
+    ).transform(transform_initial_compliance(1.0))
+    scenarios.append(always_compliant)
+
+    # Scenarios: 0% FIT Compliance
+    never_compliant = Scenario(
+        name="never_compliant", params=get_default_params()
+    ).transform(transform_initial_compliance(0.0))
+    scenarios.append(never_compliant)
+
+    # Scenarios: Delayed Onset Until Age 60
+    delayed_onset_60 = Scenario(
+        name="delayed_onset_60", params=get_default_params()
+    ).transform(transform_delayed_onset(Test.FIT, 60))
+    scenarios.append(delayed_onset_60)
+
+    # Scenarios: Delayed Onset Until Age 65
+    delayed_onset_65 = Scenario(
+        name="delayed_onset_65", params=get_default_params()
+    ).transform(transform_delayed_onset(Test.FIT, 65))
+    scenarios.append(delayed_onset_65)
+
+    # Scenario for every other year testing
+    every_two_years = Scenario(
+        name="every_two_years", params=get_default_params()
+    ).transform(transform_routine_freq(Test.FIT, 2))
+    scenarios.append(every_two_years)
+
+    # Scenario for every five years testing
+    every_five_years = Scenario(
+        name="every_five_years", params=get_default_params()
+    ).transform(transform_routine_freq(Test.FIT, 5))
+    scenarios.append(every_five_years)
+
+    # Scenario with 50% compliance every year
+    #
+    # Shorthand to create conditional compliance arrays without specifying each year.
+    # Conditional compliance arrays are length 26, one for each year of testing (50-75).
+    fifty_percent_compliance_rates = [0.5] * 26
+
+    fifty_percent_compliance = Scenario(
+        name="fifty_percent_compliance", params=get_default_params()
+    ).transform(
+        transform_conditional_compliance_rates(
+            Test.FIT,
+            ConditionalComplianceParam.PREV_COMPLIANT,
+            fifty_percent_compliance_rates,
         )
-        baseline_lower_surveillance.name = f"{fqhc}_baseline_lower_surveillance"
-        scenarios.append(baseline_lower_surveillance)
-
-        implementation_lower_surveillance = deepcopy(implementation)
-        implementation_lower_surveillance.transform(
-            transform_surveillance_frequency("polyp_mild", low_surveillance_freq_mild)
-        ).transform(
-            transform_surveillance_frequency("polyp_severe", low_surveillance_freq_severe)
-        ).transform(
-            transform_surveillance_end_age(low_surveillance_end_age)
+    )
+    scenarios.append(fifty_percent_compliance)
+
+    # Scenario with 20% compliance every year
+    twenty_percent_compliance_rates = [0.2] * 26
+
+    twenty_percent_compliance = Scenario(
+        name="twenty_percent_compliance", params=get_default_params()
+    ).transform(
+        transform_conditional_compliance_rates(
+            Test.FIT,
+            ConditionalComplianceParam.PREV_COMPLIANT,
+            twenty_percent_compliance_rates,
         )
-        implementation_lower_surveillance.name = f"{fqhc}_implementation_lower_surveillance"
-        scenarios.append(implementation_lower_surveillance)
-
+    )
+    scenarios.append(twenty_percent_compliance)
 
     return scenarios
 

crcsim/experiment/run_iteration.sh

@@ -12,7 +12,7 @@ if [ ! -d "$output_dir" ]; then
   mkdir $output_dir
 fi
 
-aws s3 cp "s3://crcsim-exp-crccp-sensitivity01/scenarios/$scenario/params.json" "./params.json" 
+aws s3 cp "s3://crcsim-exp-test-patterns/scenarios/$scenario/params.json" "./params.json" 
 
 crc-simulate \
     --npeople=$npeople \
@@ -23,5 +23,5 @@ crc-simulate \
 crc-analyze \
     --params-file=./params.json &&
 
-aws s3 cp ./results.csv "s3://crcsim-exp-crccp-sensitivity01/scenarios/$scenario/results_$iteration.csv"
-aws s3 cp ./output.csv "s3://crcsim-exp-crccp-sensitivity01/scenarios/$scenario/output_$iteration.csv"
+aws s3 cp ./results.csv "s3://crcsim-exp-test-patterns/scenarios/$scenario/results_$iteration.csv"
+aws s3 cp ./output.csv "s3://crcsim-exp-test-patterns/scenarios/$scenario/output_$iteration.csv"