|
a |
|
b/tests/test_transforms.py |
|
|
1 |
from __future__ import annotations |
|
|
2 |
|
|
|
3 |
import datetime |
|
|
4 |
|
|
|
5 |
import meds |
|
|
6 |
|
|
|
7 |
from femr.transforms import delta_encode, remove_nones |
|
|
8 |
from femr.transforms.stanford import ( |
|
|
9 |
move_billing_codes, |
|
|
10 |
move_pre_birth, |
|
|
11 |
move_to_day_end, |
|
|
12 |
move_visit_start_to_first_event_start, |
|
|
13 |
) |
|
|
14 |
|
|
|
15 |
|
|
|
16 |
def cleanup(patient): |
|
|
17 |
for event in patient["events"]: |
|
|
18 |
for measurement in event["measurements"]: |
|
|
19 |
if "metadata" not in measurement: |
|
|
20 |
measurement["metadata"] = {} |
|
|
21 |
|
|
|
22 |
for k in ("numeric_value", "text_value", "datetime_value"): |
|
|
23 |
if k not in measurement: |
|
|
24 |
measurement[k] = None |
|
|
25 |
|
|
|
26 |
if "table" not in measurement["metadata"]: |
|
|
27 |
measurement["metadata"]["table"] = None |
|
|
28 |
|
|
|
29 |
|
|
|
30 |
def test_pre_birth() -> None: |
|
|
31 |
patient = { |
|
|
32 |
"patient_id": 123, |
|
|
33 |
"events": [ |
|
|
34 |
{"time": datetime.datetime(1999, 7, 2), "measurements": [{"code": 1234}]}, |
|
|
35 |
{"time": datetime.datetime(1999, 7, 9), "measurements": [{"code": meds.birth_code}]}, |
|
|
36 |
{"time": datetime.datetime(1999, 7, 11), "measurements": [{"code": 12345}]}, |
|
|
37 |
], |
|
|
38 |
} |
|
|
39 |
|
|
|
40 |
expected = { |
|
|
41 |
"patient_id": 123, |
|
|
42 |
"events": [ |
|
|
43 |
{"time": datetime.datetime(1999, 7, 9), "measurements": [{"code": 1234}]}, |
|
|
44 |
{"time": datetime.datetime(1999, 7, 9), "measurements": [{"code": meds.birth_code}]}, |
|
|
45 |
{"time": datetime.datetime(1999, 7, 11), "measurements": [{"code": 12345}]}, |
|
|
46 |
], |
|
|
47 |
} |
|
|
48 |
|
|
|
49 |
cleanup(patient) |
|
|
50 |
cleanup(expected) |
|
|
51 |
|
|
|
52 |
assert move_pre_birth(patient) == expected |
|
|
53 |
|
|
|
54 |
|
|
|
55 |
def test_move_visit_start_ignores_other_visits() -> None: |
|
|
56 |
patient = { |
|
|
57 |
"patient_id": 123, |
|
|
58 |
"events": [ |
|
|
59 |
{ # A non-visit event with no explicit start time |
|
|
60 |
"time": datetime.datetime(1999, 7, 2), |
|
|
61 |
"measurements": [{"code": 1234, "metadata": {"visit_id": 9999}}], |
|
|
62 |
}, |
|
|
63 |
{ # A visit event with just date specified |
|
|
64 |
"time": datetime.datetime(1999, 7, 2), |
|
|
65 |
"measurements": [ |
|
|
66 |
{ |
|
|
67 |
"code": 4567, |
|
|
68 |
"metadata": { |
|
|
69 |
"table": "visit", |
|
|
70 |
"visit_id": 9999, |
|
|
71 |
}, |
|
|
72 |
} |
|
|
73 |
], |
|
|
74 |
}, |
|
|
75 |
{ # A non-visit event from a separate visit ID |
|
|
76 |
"time": datetime.datetime(1999, 7, 2, 11), |
|
|
77 |
"measurements": [{"code": 2345, "metadata": {"visit_id": 8888}}], |
|
|
78 |
}, |
|
|
79 |
{ # First recorded non-visit event for visit ID 9999 |
|
|
80 |
"time": datetime.datetime(1999, 7, 2, 12), |
|
|
81 |
"measurements": [{"code": 3456, "metadata": {"visit_id": 9999}}], |
|
|
82 |
}, |
|
|
83 |
], |
|
|
84 |
} |
|
|
85 |
|
|
|
86 |
# Note that events are implicitly sorted first by start time, then by code: |
|
|
87 |
# https://github.com/som-shahlab/femr/blob/main/src/femr/__init__.py#L69 |
|
|
88 |
expected = { |
|
|
89 |
"patient_id": 123, |
|
|
90 |
"events": [ |
|
|
91 |
{ # A non-visit event with no explicit start time |
|
|
92 |
"time": datetime.datetime(1999, 7, 2), |
|
|
93 |
"measurements": [{"code": 1234, "metadata": {"visit_id": 9999}}], |
|
|
94 |
}, |
|
|
95 |
{ # A non-visit event from a separate visit ID |
|
|
96 |
"time": datetime.datetime(1999, 7, 2, 11), |
|
|
97 |
"measurements": [{"code": 2345, "metadata": {"visit_id": 8888}}], |
|
|
98 |
}, |
|
|
99 |
{ # Now visit event has date and time specified |
|
|
100 |
"time": datetime.datetime(1999, 7, 2, 12), |
|
|
101 |
"measurements": [ |
|
|
102 |
{ |
|
|
103 |
"code": 4567, |
|
|
104 |
"metadata": { |
|
|
105 |
"table": "visit", |
|
|
106 |
"visit_id": 9999, |
|
|
107 |
}, |
|
|
108 |
} |
|
|
109 |
], |
|
|
110 |
}, |
|
|
111 |
{ # First recorded non-visit event for visit ID 9999 |
|
|
112 |
"time": datetime.datetime(1999, 7, 2, 12), |
|
|
113 |
"measurements": [{"code": 3456, "metadata": {"visit_id": 9999}}], |
|
|
114 |
}, |
|
|
115 |
], |
|
|
116 |
} |
|
|
117 |
|
|
|
118 |
cleanup(patient) |
|
|
119 |
cleanup(expected) |
|
|
120 |
|
|
|
121 |
assert move_visit_start_to_first_event_start(patient) == expected |
|
|
122 |
|
|
|
123 |
|
|
|
124 |
def test_move_visit_start_minute_after_midnight() -> None: |
|
|
125 |
patient = { |
|
|
126 |
"patient_id": 123, |
|
|
127 |
"events": [ |
|
|
128 |
{ |
|
|
129 |
"time": datetime.datetime(1999, 7, 2), |
|
|
130 |
"measurements": [ |
|
|
131 |
{"code": 3456, "metadata": {"visit_id": 9999, "table": "visit"}}, |
|
|
132 |
{"code": 1234, "metadata": {"visit_id": 9999}}, |
|
|
133 |
], |
|
|
134 |
}, |
|
|
135 |
{ |
|
|
136 |
"time": datetime.datetime(1999, 7, 2, 0, 1), |
|
|
137 |
"measurements": [{"code": 2345, "metadata": {"visit_id": 9999}}], |
|
|
138 |
}, |
|
|
139 |
{ |
|
|
140 |
"time": datetime.datetime(1999, 7, 2, 12), |
|
|
141 |
"measurements": [{"code": 4567, "metadata": {"visit_id": 9999}}], |
|
|
142 |
}, |
|
|
143 |
], |
|
|
144 |
} |
|
|
145 |
|
|
|
146 |
expected = { |
|
|
147 |
"patient_id": 123, |
|
|
148 |
"events": [ |
|
|
149 |
{ |
|
|
150 |
"time": datetime.datetime(1999, 7, 2), |
|
|
151 |
"measurements": [{"code": 1234, "metadata": {"visit_id": 9999}}], |
|
|
152 |
}, |
|
|
153 |
{ |
|
|
154 |
"time": datetime.datetime(1999, 7, 2, 0, 1), |
|
|
155 |
"measurements": [{"code": 3456, "metadata": {"visit_id": 9999, "table": "visit"}}], |
|
|
156 |
}, |
|
|
157 |
{ |
|
|
158 |
"time": datetime.datetime(1999, 7, 2, 0, 1), |
|
|
159 |
"measurements": [{"code": 2345, "metadata": {"visit_id": 9999}}], |
|
|
160 |
}, |
|
|
161 |
{ |
|
|
162 |
"time": datetime.datetime(1999, 7, 2, 12), |
|
|
163 |
"measurements": [{"code": 4567, "metadata": {"visit_id": 9999}}], |
|
|
164 |
}, |
|
|
165 |
], |
|
|
166 |
} |
|
|
167 |
|
|
|
168 |
cleanup(patient) |
|
|
169 |
cleanup(expected) |
|
|
170 |
|
|
|
171 |
assert move_visit_start_to_first_event_start(patient) == expected |
|
|
172 |
|
|
|
173 |
|
|
|
174 |
def test_move_visit_start_doesnt_move_without_event() -> None: |
|
|
175 |
patient = { |
|
|
176 |
"patient_id": 123, |
|
|
177 |
"events": [ |
|
|
178 |
{ |
|
|
179 |
"time": datetime.datetime(1999, 7, 2), |
|
|
180 |
"measurements": [ |
|
|
181 |
{"code": 1234, "metadata": {"visit_id": 9999}}, |
|
|
182 |
{"code": 3456, "metadata": {"visit_id": 9999, "table": "visit"}}, |
|
|
183 |
{"code": 2345, "metadata": {"visit_id": 9999}}, |
|
|
184 |
], |
|
|
185 |
}, |
|
|
186 |
], |
|
|
187 |
} |
|
|
188 |
|
|
|
189 |
# None of the non-visit events have start time > '00:00:00' so visit event |
|
|
190 |
# start time is unchanged, though order changes based on code under resort. |
|
|
191 |
expected = { |
|
|
192 |
"patient_id": 123, |
|
|
193 |
"events": [ |
|
|
194 |
{ |
|
|
195 |
"time": datetime.datetime(1999, 7, 2), |
|
|
196 |
"measurements": [ |
|
|
197 |
{"code": 1234, "metadata": {"visit_id": 9999}}, |
|
|
198 |
{"code": 3456, "metadata": {"visit_id": 9999, "table": "visit"}}, |
|
|
199 |
{"code": 2345, "metadata": {"visit_id": 9999}}, |
|
|
200 |
], |
|
|
201 |
} |
|
|
202 |
], |
|
|
203 |
} |
|
|
204 |
|
|
|
205 |
cleanup(patient) |
|
|
206 |
cleanup(expected) |
|
|
207 |
|
|
|
208 |
assert move_visit_start_to_first_event_start(patient) == expected |
|
|
209 |
|
|
|
210 |
|
|
|
211 |
def test_move_to_day_end() -> None: |
|
|
212 |
patient = { |
|
|
213 |
"patient_id": 123, |
|
|
214 |
"events": [ |
|
|
215 |
{"time": datetime.datetime(1999, 7, 2), "measurements": [{"code": 1234}]}, |
|
|
216 |
{"time": datetime.datetime(1999, 7, 2, 12), "measurements": [{"code": 4321}]}, |
|
|
217 |
{"time": datetime.datetime(1999, 7, 9), "measurements": [{"code": meds.birth_code}]}, |
|
|
218 |
], |
|
|
219 |
} |
|
|
220 |
|
|
|
221 |
expected = { |
|
|
222 |
"patient_id": 123, |
|
|
223 |
"events": [ |
|
|
224 |
{"time": datetime.datetime(1999, 7, 2, 12), "measurements": [{"code": 4321}]}, |
|
|
225 |
{"time": datetime.datetime(1999, 7, 2, 23, 59), "measurements": [{"code": 1234}]}, |
|
|
226 |
{"time": datetime.datetime(1999, 7, 9, 23, 59), "measurements": [{"code": meds.birth_code}]}, |
|
|
227 |
], |
|
|
228 |
} |
|
|
229 |
|
|
|
230 |
cleanup(patient) |
|
|
231 |
cleanup(expected) |
|
|
232 |
|
|
|
233 |
print(move_to_day_end(patient)) |
|
|
234 |
print(expected) |
|
|
235 |
|
|
|
236 |
assert move_to_day_end(patient) == expected |
|
|
237 |
|
|
|
238 |
|
|
|
239 |
def test_remove_nones() -> None: |
|
|
240 |
patient = { |
|
|
241 |
"patient_id": 123, |
|
|
242 |
"events": [ |
|
|
243 |
{"time": datetime.datetime(1999, 7, 2), "measurements": [{"code": 1234}]}, # No value, to be removed |
|
|
244 |
{"time": datetime.datetime(1999, 7, 2, 12), "measurements": [{"code": 1234, "numeric_value": 3}]}, |
|
|
245 |
{"time": datetime.datetime(1999, 7, 9), "measurements": [{"code": meds.birth_code}]}, |
|
|
246 |
], |
|
|
247 |
} |
|
|
248 |
|
|
|
249 |
expected = { |
|
|
250 |
"patient_id": 123, |
|
|
251 |
"events": [ |
|
|
252 |
{"time": datetime.datetime(1999, 7, 2, 12), "measurements": [{"code": 1234, "numeric_value": 3}]}, |
|
|
253 |
{"time": datetime.datetime(1999, 7, 9), "measurements": [{"code": meds.birth_code}]}, |
|
|
254 |
], |
|
|
255 |
} |
|
|
256 |
|
|
|
257 |
cleanup(patient) |
|
|
258 |
cleanup(expected) |
|
|
259 |
|
|
|
260 |
assert remove_nones(patient) == expected |
|
|
261 |
|
|
|
262 |
|
|
|
263 |
def test_delta_encode() -> None: |
|
|
264 |
patient = { |
|
|
265 |
"patient_id": 123, |
|
|
266 |
"events": [ |
|
|
267 |
{"time": datetime.datetime(1999, 7, 2), "measurements": [{"code": 1234}]}, |
|
|
268 |
{"time": datetime.datetime(1999, 7, 2), "measurements": [{"code": 1234}]}, |
|
|
269 |
{"time": datetime.datetime(1999, 7, 2, 12), "measurements": [{"code": 1234, "numeric_value": 3}]}, |
|
|
270 |
{"time": datetime.datetime(1999, 7, 2, 14), "measurements": [{"code": 1234, "numeric_value": 3}]}, |
|
|
271 |
{"time": datetime.datetime(1999, 7, 2, 19), "measurements": [{"code": 1234, "numeric_value": 5}]}, |
|
|
272 |
{"time": datetime.datetime(1999, 7, 2, 20), "measurements": [{"code": 1234, "numeric_value": 3}]}, |
|
|
273 |
], |
|
|
274 |
} |
|
|
275 |
|
|
|
276 |
expected = { |
|
|
277 |
"patient_id": 123, |
|
|
278 |
"events": [ |
|
|
279 |
{"time": datetime.datetime(1999, 7, 2), "measurements": [{"code": 1234}]}, |
|
|
280 |
{"time": datetime.datetime(1999, 7, 2, 12), "measurements": [{"code": 1234, "numeric_value": 3}]}, |
|
|
281 |
{"time": datetime.datetime(1999, 7, 2, 19), "measurements": [{"code": 1234, "numeric_value": 5}]}, |
|
|
282 |
{"time": datetime.datetime(1999, 7, 2, 20), "measurements": [{"code": 1234, "numeric_value": 3}]}, |
|
|
283 |
], |
|
|
284 |
} |
|
|
285 |
|
|
|
286 |
cleanup(patient) |
|
|
287 |
cleanup(expected) |
|
|
288 |
|
|
|
289 |
assert delta_encode(patient) == expected |
|
|
290 |
|
|
|
291 |
|
|
|
292 |
def test_move_billing_codes() -> None: |
|
|
293 |
patient = { |
|
|
294 |
"patient_id": 123, |
|
|
295 |
"events": [ |
|
|
296 |
{ |
|
|
297 |
"time": datetime.datetime(1999, 7, 2), |
|
|
298 |
"measurements": [ |
|
|
299 |
{ |
|
|
300 |
"code": 1234, |
|
|
301 |
"metadata": { |
|
|
302 |
"end": datetime.datetime(1999, 7, 20), |
|
|
303 |
"visit_id": 10, |
|
|
304 |
"clarity_table": "lpch_pat_enc", |
|
|
305 |
}, |
|
|
306 |
} |
|
|
307 |
], |
|
|
308 |
}, |
|
|
309 |
{ |
|
|
310 |
"time": datetime.datetime(1999, 7, 9), |
|
|
311 |
"measurements": [ |
|
|
312 |
{ |
|
|
313 |
"code": meds.birth_code, |
|
|
314 |
"metadata": { |
|
|
315 |
"visit_id": 10, |
|
|
316 |
"clarity_table": "lpch_pat_enc_dx", |
|
|
317 |
}, |
|
|
318 |
} |
|
|
319 |
], |
|
|
320 |
}, |
|
|
321 |
{ |
|
|
322 |
"time": datetime.datetime(1999, 7, 10), |
|
|
323 |
"measurements": [ |
|
|
324 |
{ |
|
|
325 |
"code": 42165, |
|
|
326 |
"metadata": { |
|
|
327 |
"visit_id": 10, |
|
|
328 |
"clarity_table": "shc_pat_enc_dx", |
|
|
329 |
}, |
|
|
330 |
} |
|
|
331 |
], |
|
|
332 |
}, |
|
|
333 |
{ |
|
|
334 |
"time": datetime.datetime(1999, 7, 11), |
|
|
335 |
"measurements": [ |
|
|
336 |
{ |
|
|
337 |
"code": 12345, |
|
|
338 |
"metadata": { |
|
|
339 |
"visit_id": 10, |
|
|
340 |
}, |
|
|
341 |
} |
|
|
342 |
], |
|
|
343 |
}, |
|
|
344 |
{ |
|
|
345 |
"time": datetime.datetime(1999, 7, 13), |
|
|
346 |
"measurements": [ |
|
|
347 |
{ |
|
|
348 |
"code": 123, |
|
|
349 |
"metadata": { |
|
|
350 |
"visit_id": 11, |
|
|
351 |
}, |
|
|
352 |
} |
|
|
353 |
], |
|
|
354 |
}, |
|
|
355 |
], |
|
|
356 |
} |
|
|
357 |
|
|
|
358 |
expected = { |
|
|
359 |
"patient_id": 123, |
|
|
360 |
"events": [ |
|
|
361 |
{ |
|
|
362 |
"time": datetime.datetime(1999, 7, 2), |
|
|
363 |
"measurements": [ |
|
|
364 |
{ |
|
|
365 |
"code": 1234, |
|
|
366 |
"metadata": { |
|
|
367 |
"end": datetime.datetime(1999, 7, 20), |
|
|
368 |
"visit_id": 10, |
|
|
369 |
"clarity_table": "lpch_pat_enc", |
|
|
370 |
}, |
|
|
371 |
} |
|
|
372 |
], |
|
|
373 |
}, |
|
|
374 |
{ |
|
|
375 |
"time": datetime.datetime(1999, 7, 11), |
|
|
376 |
"measurements": [ |
|
|
377 |
{ |
|
|
378 |
"code": 12345, |
|
|
379 |
"metadata": { |
|
|
380 |
"visit_id": 10, |
|
|
381 |
}, |
|
|
382 |
} |
|
|
383 |
], |
|
|
384 |
}, |
|
|
385 |
{ |
|
|
386 |
"time": datetime.datetime(1999, 7, 13), |
|
|
387 |
"measurements": [ |
|
|
388 |
{ |
|
|
389 |
"code": 123, |
|
|
390 |
"metadata": { |
|
|
391 |
"visit_id": 11, |
|
|
392 |
}, |
|
|
393 |
} |
|
|
394 |
], |
|
|
395 |
}, |
|
|
396 |
{ |
|
|
397 |
"time": datetime.datetime(1999, 7, 20), |
|
|
398 |
"measurements": [ |
|
|
399 |
{ |
|
|
400 |
"code": meds.birth_code, |
|
|
401 |
"metadata": { |
|
|
402 |
"visit_id": 10, |
|
|
403 |
"clarity_table": "lpch_pat_enc_dx", |
|
|
404 |
}, |
|
|
405 |
} |
|
|
406 |
], |
|
|
407 |
}, |
|
|
408 |
{ |
|
|
409 |
"time": datetime.datetime(1999, 7, 20), |
|
|
410 |
"measurements": [ |
|
|
411 |
{ |
|
|
412 |
"code": 42165, |
|
|
413 |
"metadata": { |
|
|
414 |
"visit_id": 10, |
|
|
415 |
"clarity_table": "shc_pat_enc_dx", |
|
|
416 |
}, |
|
|
417 |
} |
|
|
418 |
], |
|
|
419 |
}, |
|
|
420 |
], |
|
|
421 |
} |
|
|
422 |
|
|
|
423 |
cleanup(patient) |
|
|
424 |
cleanup(expected) |
|
|
425 |
|
|
|
426 |
print(move_billing_codes(patient)) |
|
|
427 |
print(expected) |
|
|
428 |
|
|
|
429 |
assert move_billing_codes(patient) == expected |