File size: 13,684 Bytes
128403f
 
f137b36
 
 
128403f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f137b36
128403f
 
 
f137b36
128403f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
772c682
128403f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f137b36
128403f
 
 
 
 
 
 
 
f137b36
128403f
 
 
 
f137b36
128403f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
772c682
128403f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f137b36
 
128403f
 
 
 
 
 
 
 
 
 
 
 
 
f137b36
128403f
 
 
 
 
 
 
 
 
 
 
772c682
128403f
 
 
 
 
 
 
 
 
 
 
d6b88c2
128403f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
772c682
128403f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
"""
Gradio user interface for AIM-CU

Disclaimer:
This software and documentation was developed at the Food and Drug Administration (FDA) by employees of the Federal Government in the course of their official duties. Pursuant to Title 17, Section 105 of the United States Code, this work is not subject to copyright protection and is in the public domain. Permission is hereby granted, free of charge, to any person obtaining a copy of the Software, to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, or sell copies of the Software or derivatives, and to permit persons to whom the Software is furnished to do so. FDA assumes no responsibility whatsoever for use by other parties of the Software, its source code, documentation or compiled executables, and makes no guarantees, expressed or implied, about its quality, reliability, or any other characteristic. Further, use of this code in no way implies endorsement by the FDA or confers any advantage in regulatory decisions. Although this software can be redistributed and/or modified freely, we ask that any derivative works bear some notice that they are derived from it, and any modified versions bear some notice that they have been modified.
"""

import os
import sys
import pandas as pd
import gradio as gr
import tomli
from cusum import CUSUM
from ARLTheoretical import get_ref_value, get_ref_value_k, get_ARL_1, get_ARL_1_h_mu1_k
from utils import (
    populate_summary_table_ARL0_k,
    populate_summary_table_ARL1_k,
)
import great_tables as gt
import plotly.graph_objects as go


def set_init_days(
    file_csv_metric: gr.File, init_days: str
) -> tuple[float, float, go.Figure]:
    """
    Set number of baseline observations and get in-control mean and standard deviation.

    Args:
        file_csv_metric (gr.File): CSV file with metric data
        init_days (str): Number of baseline observations to calculate in-control mean and standard deviation

    Returns:
        tuple[float, float, go.Figure]: In-control mean and standard deviation, and observation data plot.
    """
    init_days = int(init_days)

    data_csv_metric = pd.read_csv(file_csv_metric.name)
    obj_cusum.set_df_metric_csv(data_csv_metric)

    obj_cusum.set_init_stats(init_days=init_days)

    return (
        "{:.2f}".format(obj_cusum.in_mu),
        "{:.2f}".format(obj_cusum.in_std),
        obj_cusum.plot_input_metric_plotly_raw(),
    )


def populate_table(h: str) -> tuple[gt.GT, gt.GT]:
    """
    Populate tables for ARL0 and ARL1 given the value of h

    Args:
        h (str): Normalized threshold.

    Returns:
        tuple[gt.GT, gt.GT]: Table for ARL0 and k in HTML format; table for ARL1 and k in HTML format.
    """
    h = float(h)

    summary_table_df_ARL0_k, dict_ARL0_k = get_ref_value(
        h=h,
        list_ARL_0=config["params_cusum"]["list_ARL_0"],
    )
    summary_table_df_ARL0_k = summary_table_df_ARL0_k.applymap(
        lambda x: "{:.2f}".format(x) if isinstance(x, (float)) else x
    )

    summary_table_df_ARL1_k = get_ARL_1(
        h=h,
        shift_in_mean=config["params_cusum"]["shift_in_mean"],
        dict_ARL0_k=dict_ARL0_k,
    )
    summary_table_df_ARL1_k = summary_table_df_ARL1_k.applymap(
        lambda x: "{:.2f}".format(x) if isinstance(x, (float)) else x
    )

    return populate_summary_table_ARL0_k(
        summary_table_df_ARL0_k, h
    ), populate_summary_table_ARL1_k(summary_table_df_ARL1_k, dict_ARL0_k, h)


def calculate_reference_value_k(h: str, arl_0: str) -> float:
    """
    Gets the reference value for given h and ARL_0.

    Args:
        h (str): Normalized threshold.
        arl_0 (str): ARL0 value.

    Returns:
        float: Normalized reference value k.
    """
    h = float(h)
    arl_0 = float(arl_0)

    k = get_ref_value_k(h=h, ARL_0=arl_0)
    k = "{:.2f}".format(k)

    return k, k, k


def calculate_arl1_h_k_mu1(h: str, k: str, mu1: str) -> float:
    """
    Get the ARL_1 with given Shift in Mean (mu1) and k.

    Args:
        h (str): Normalized threshold.
        k (str): Normalized reference value.
        mu1 (str): Intended shift in mean.

    Returns:
        float: Detection delay (ARL1).
    """
    h = float(h)
    k = float(k)
    mu1 = float(mu1)

    arl_1 = get_ARL_1_h_mu1_k(h=h, k=k, mu1=mu1)
    arl_1 = "{:.2f}".format(arl_1)

    return arl_1


def populate_cusum_plots(
    ref_value: str, normalized_threshold: str
) -> tuple[go.Figure, go.Figure]:
    """
    Populate CUSUM plots

    Args:
        ref_value (str): Normalized reference value for detecting a unit standard deviation change in mean of the process.
        normalized_threshold (str): Normalized threshold.

    Returns:
        tuple[go.Figure, go.Figure]: Scatter plot as Plotly graph object; CUSUM plot using Plotly graph object.
    """
    ref_value = float(ref_value)
    normalized_threshold = float(normalized_threshold)

    obj_cusum.change_detection(
        normalized_ref_value=ref_value,
        normalized_threshold=normalized_threshold,
    )

    return (obj_cusum.plot_input_metric_plotly(), obj_cusum.plot_cusum_plotly())


with gr.Blocks(
    theme=gr.themes.Base(
        neutral_hue=gr.themes.Color(
            c50="#e5f1f8",
            c100="#e5f1f8",
            c200="#cce4f1",
            c300="#b2d7ea",
            c400="#7fbddc",
            c500="#4ca3ce",
            c600="#007cba",
            c700="#006394",
            c800="#004a6f",
            c900="#00314a",
            c950="#001825",
        ),
    )
) as demo:
    gr.Markdown(f"""
                # AIM-CU
                ## AIM-CU: A statistical tool for AI monitoring using cumulative sum (AIM-CU).
                """)  # noqa: F541

    gr.Markdown(f"""
                ### AIM-CU Input:
                AI output (e.g. metrics such as Accuracy, F1-score, Sensitivity etc.)
                """)  # noqa: F541

    with gr.Row():
        with gr.Column():
            gr.Markdown(f"""
                        ### Initialization:
                        """)  # noqa: F541

            gr.Markdown(f"""
                ### Upload the AI output.
                """)  # noqa: F541

            # load the CSV file with specifities across days
            csv_file_metric = gr.File(
                label="Upload the AI output (CSV file)",
            )

            with gr.Row():
                with gr.Column():
                    init_days = gr.Textbox(label="Number of baseline observations", placeholder="30")
                with gr.Column():
                    button_calculate_incontrol_params = gr.Button(
                        "Calculate parameters"
                    )

            with gr.Row():
                with gr.Column():
                    in_control_mean = gr.Textbox(
                        label="In-control mean", interactive=False
                    )
                with gr.Column():
                    in_control_std = gr.Textbox(
                        label="In-control standard deviation", interactive=False
                    )

            plot_observation_data = gr.Plot(label="AI output", visible=False)

            gr.Markdown(f"""
                        Parameter choices for detecting change and detection delay estimates (theoretical calculations).
                        """)  # noqa: F541

            gr.Markdown(f"""
                ### Enter h value:
                """)  # noqa: F541

            h_phase1 = gr.Textbox(
                label="h value =",
                placeholder="h = normalized threshold, default = 4. Range: between 4 and 5 ([4, 5])",
                value="3",
                autofocus=True,
            )

            dataframe_gt_ref_value = gr.HTML(
                label="Reference Values for an intended ARL0 with normalized threshold h",
                show_label=True,
                visible=False,
            )

            gr.Markdown(f"""
                ### Calculate reference value k for a specific value for ARL<sub>0</sub>:
                """)  # noqa: F541

            with gr.Row():
                arl_0 = gr.Textbox(
                    label="ARL_0 value =", placeholder="ARL_0", value="100"
                )

                button_calculate_k = gr.Button("Calculate k")

                output_k = gr.Textbox(label="Calculated k =", visible=False)

            dataframe_gt_ARL0 = gr.HTML(
                label="Estimate of steady state ARL (ARL_1 based on the computed reference values and intended zero-state ARL (ARL_0) with normalized threshold h)",
                show_label=True,
                visible=False,
            )

            gr.Markdown(f"""
                ### Calculate ARL<sub>1</sub> for reference value h, value k and shift in mean:
                """)  # noqa: F541

            with gr.Row():
                k_phase1 = gr.Textbox(
                    label="k value =", placeholder="k", value="0.2996"
                )
                mu1 = gr.Textbox(
                    label="Shift in mean value =",
                    placeholder="Shift in mean value",
                    value="1.2",
                )

                button_calculate_ARL_1 = gr.Button("Calculate ARL_1")

                output_ARL_1 = gr.Textbox(label="Calculated ARL_1 =", visible=False)

            button_populate_table = gr.Button(
                "Populate Reference Values and ARL_1 tables for the given h value"
            )

            gr.Markdown(f"""
                ### Workflow:
                Phase I:
                - Upload the AI output (CSV file).
                - Enter number of baseline observations.
                - Calculate parameters.
                - Check parameter choices in Phase 1, for Phase 2. (optional)
                
                Phase II:
                - Enter h and k values.
                - Get CUSUM plots.
                """)  # noqa: F541

            table_param_description = gr.Dataframe(
                value=pd.read_csv("../../assets/params.csv")
            )
        with gr.Column():
            gr.Markdown(f"""
                        ### Monitoring:
                        Performance drift detection plots, pre- and post-change distribution with respect to the performance drift detected.
                        """)  # noqa: F541

            gr.Markdown(f"""
                ### Enter required values:
                """)  # noqa: F541

            with gr.Row():
                h_phase2 = gr.Textbox(
                    label="h value =",
                    placeholder="normalized threshold, default = 4. Range: between 4 and 5 ([4, 5])",
                    value="3",
                )

                k_phase2 = gr.Textbox(
                    label="k value =",
                    placeholder="normalized reference value, default = 0.5",
                    value="0.5",
                )

            button_csv_metric = gr.Button("Show CUSUM plots")

            plot_avg_metric = gr.Plot(
                label="AI model performance",
                visible=False,
            )
            plot_cusum_chart = gr.Plot(label="CUSUM Chart", visible=False)

    button_calculate_incontrol_params.click(
        fn=set_init_days,
        inputs=[csv_file_metric, init_days],
        outputs=[in_control_mean, in_control_std, plot_observation_data],
    )

    button_calculate_incontrol_params.click(
        fn=lambda: gr.update(visible=True), inputs=[], outputs=plot_observation_data
    )

    # Get the CSV file and populate tables
    button_populate_table.click(
        fn=populate_table,
        inputs=[h_phase1],
        outputs=[dataframe_gt_ref_value, dataframe_gt_ARL0],
    )
    button_populate_table.click(
        fn=lambda: gr.update(visible=True), inputs=[], outputs=dataframe_gt_ref_value
    )
    button_populate_table.click(
        fn=lambda: gr.update(visible=True), inputs=[], outputs=dataframe_gt_ARL0
    )

    # Calculate specific k for ARL_0
    button_calculate_k.click(
        fn=calculate_reference_value_k, inputs=[h_phase1, arl_0], outputs=[output_k, k_phase1, k_phase2]
    )
    button_calculate_k.click(
        fn=lambda: gr.update(visible=True), inputs=[], outputs=output_k
    )

    # Calculate specific ARL_1 for value h, value k and shift in mean
    button_calculate_ARL_1.click(
        fn=calculate_arl1_h_k_mu1,
        inputs=[h_phase1, k_phase1, mu1],
        outputs=[output_ARL_1],
    )
    button_calculate_ARL_1.click(
        fn=lambda: gr.update(visible=True), inputs=[], outputs=output_ARL_1
    )

    # Get the CSV file and populate plots
    button_csv_metric.click(
        fn=populate_cusum_plots,
        inputs=[k_phase2, h_phase2],
        outputs=[plot_avg_metric, plot_cusum_chart],
    )

    button_csv_metric.click(
        fn=lambda: gr.update(visible=True), inputs=[], outputs=plot_avg_metric
    )
    button_csv_metric.click(
        fn=lambda: gr.update(visible=True), inputs=[], outputs=plot_cusum_chart
    )


try:
    path_file_config = os.path.abspath("../../config/config.toml")

    with open(path_file_config, "rb") as file_config:
        config = tomli.load(file_config)
except FileNotFoundError:
    print("Error: config.toml not found at", path_file_config)
    sys.exit(1)


if config["control"]["save_figure"] == "true":
    path_check = os.path.abspath(
        os.path.join("../../", config["path_output"]["path_figure"])
    )
    if not os.path.exists(path_check):
        os.mkdir(path_check)
        print("Created", path_check)

obj_cusum = CUSUM()
obj_cusum.initialize()

demo.launch(server_name="0.0.0.0", server_port=7860)