import pandas as pd # import seaborn as sns # import numpy as np from scipy.stats import pearsonr, spearmanr import plotly_express as px import plotly import plotly.graph_objects as go import re # import statsmodels # import math

df = pd.read_csv('rawdata.csv')

PGY_LEVEL = 'pgy_level' OPERATION = 'operation' SCORE = 'score' PGY_DICT = { 'PGY1' : 1, 'PGY2' : 2, 'PGY3' : 3, 'Research Year Resident' : 4, 'PGY4' : 5, 'PGY5' : 6 } PGY_REVERSE_DICT = { 1: 'PGY1', 2: 'PGY2', 3: 'PGY3', 4: 'Research', 5: 'PGY4', 6: 'PGY5' } PGY_DOUBLE_REVERESE_DICT = { 'PGY1' : 1, 'PGY2' : 2, 'PGY3' : 3, 'Research' : 4, 'PGY4' : 5, 'PGY5' : 6 } PORT_SCORE_DICT = { 'Observation Only' : 1, 'Direct Supervision' : 2, 'Indirect Supervision' : 3, 'Unsupervised Practice' : 4, 'Supervising Others' : 5 }

THRESHOLD = 4

df.replace({PGY_LEVEL: PGY_DICT}, inplace=True) proc_cols = list(new_cols.values())[2:] for col in proc_cols: df.replace({col : PORT_SCORE_DICT}, inplace=True) df

## Set a Color R1 = '69' B1 = '145' G1 = '226' ## Set a Color R2 = '15' B2 = '149' G2 = '215' ## Set a Color R3 = '174' B3 = '214' G3 = '234' ## Set a Color R4 = '243' B4 = '243' G4 = '243' ## Set a Color R5 = '190' B5 = '190' G5 = '190' darkblue = 'rgba(' + R2 + ', ' + B2 + ', ' + G2 + ', 1.0)' lightblue = 'rgba(' + R3 + ', ' + B3 + ', ' + G3 + ', 1.0)' lightgray = 'rgba(' + R4 + ', ' + B4 + ', ' + G4 + ', 1.0)' darkgray = 'rgba(' + R5 + ', ' + B5 + ', ' + G5 + ', 1.0)' # solid_color = 'rgba(' + R + ', ' + B + ', ' + G + ', 1.0)' # bg_color = 'rgba(' + R + ', ' + B + ', ' + G + ', 0.2)' no_color = 'rgba(' + R1 + ', ' + B1 + ', ' + G1 + ', 0.0)'

df_response = df.groupby(PGY_LEVEL).agg( responses = ('Timestamp', 'count') ) df_response.reset_index(inplace=True) df_response.replace({PGY_LEVEL: PGY_REVERSE_DICT}, inplace=True) df_response['total'] = [7, 7, 8, 11, 6, 7] df_response['percent'] = ((df_response['responses']/df_response['total']).round(2)*100).astype(int).astype(str) + '%' df_response

df_melt = pd.melt(df, id_vars=PGY_LEVEL, value_vars=df.columns[2:], var_name='operation', value_name='score') df_melt.sort_values(by=[OPERATION, PGY_LEVEL], inplace=True) # print(df) df_melt = df_melt.dropna() df_melt = df_melt.reset_index(drop=True) df_melt # df.to_csv('rawscores.csv')

dict = {OPERATION:[],'correlation':[],'pvalue':[]} for name, group in df_melt.groupby('operation'): corr, pvalue = spearmanr(group[PGY_LEVEL], group[SCORE]) dict['operation'].append(name) dict['correlation'].append(corr) dict['pvalue'].append(pvalue) corr_df = pd.DataFrame(dict) corr_df['correlation'] = corr_df['correlation'].round(2) corr_df['p-value'] = corr_df.apply(lambda x: '{:.2e}'.format(x['pvalue']), axis=1) # corr_df['p-value'] = corr_df.apply(lambda x: round(x['pvalue'], 3), axis=1) # corr_df.drop('pvalue', inplace=True, axis=1) corr_df

df_count_per_score = df_melt.groupby([OPERATION, PGY_LEVEL, SCORE]).agg( count_per_score = ('score', 'count') ) df_count_per_score.reset_index(inplace=True) df_count_per_score df_scatter = pd.merge(df_melt, df_count_per_score, on=[OPERATION, PGY_LEVEL, SCORE]) df_scatter

fig = px.scatter(df_scatter, x=PGY_LEVEL, y=SCORE, trendline='ols', size='count_per_score', facet_col=OPERATION, facet_col_wrap=4, facet_row_spacing=0.04, facet_col_spacing=0.02, color_discrete_sequence=[darkblue], opacity=1 ) fig.update_layout(title = 'Correlation between PGY Level and Operative Comfort', xaxis_title = 'PGY Level', yaxis_title = 'score', width = 1200, height = 1600, font_family='Inter', plot_bgcolor=lightgray) fig.update_traces(marker_line_width=0) fig.for_each_annotation(lambda a: a.update(text=a.text.split("=")[1])) fig.update_xaxes(tickvals=[1,2,3,4,5,6], ticktext=['1', '2', '3', 'R', '4', '5']), fig.update_yaxes(tickvals=[1, 2, 3, 4, 5], ticktext=['OB', 'DS', 'IS', 'UP', 'SO']) fig.show() fig.write_image('fig_scatter.svg')

df_scatter

df_stats = df_melt.groupby([OPERATION, PGY_LEVEL]).agg( num_residents = ('score', 'count'), min = ('score', 'min'), mean = ('score', 'mean'), median = ('score', 'median'), max = ('score', 'max'), stdev = ('score', 'std') ) df_stats.reset_index(inplace=True) df_stats['mean'] = df_stats['mean'].round(2) df_stats['stdev'] = df_stats['stdev'].round(2) df_stats['min'] = df_stats['min'].astype(int) df_stats['max'] = df_stats['max'].astype(int) df_stats['range'] = df_stats['max']-df_stats['min'] df_stats.sort_values(by=['operation', 'pgy_level'], ascending=False, inplace=True) df_stats.replace({PGY_LEVEL: PGY_REVERSE_DICT}, inplace=True) df_stats

df_stats = df_melt.groupby([OPERATION, PGY_LEVEL]).agg( num_residents = ('score', 'count'), min = ('score', 'min'), # mean = ('score', 'mean'), median = ('score', 'median'), max = ('score', 'max'), # stdev = ('score', 'std') ) df_stats.reset_index(inplace=True) # df_stats['mean'] = df_stats['mean'].round(2) # df_stats['stdev'] = df_stats['stdev'].round(2) # df_stats['min'] = df_stats['min'].astype(int) # df_stats['max'] = df_stats['max'].astype(int) df_stats['range'] = df_stats['max']-df_stats['min'] df_stats.sort_values(by=['operation', 'pgy_level'], ascending=True, inplace=True) df_stats.replace({PGY_LEVEL: PGY_REVERSE_DICT}, inplace=True) df_stats

# fig = px.line(df_stats, x=PGY_LEVEL, y='min', facet_col=OPERATION, facet_col_wrap=4, # facet_row_spacing=0.04, # default is 0.07 when facet_col_wrap is used # facet_col_spacing=0.02, # default is 0.03` # ) # fig.update_traces(line_color=no_color) # for t in fig.data: # row = t['xaxis'] # col = t['yaxis'] # proc = t['hovertemplate'].strip('operation=') # proc = re.sub('(<br>.+)', '', proc) # dff = df_stats.loc[df_stats['operation'] == proc] # fig.add_trace(go.Scatter(x=dff[PGY_LEVEL], # y=dff['max'], # xaxis=row, # yaxis=col, # fill='tonexty', # fillcolor=lightblue, # line_color=no_color)) # fig.add_trace(go.Scatter(x=dff[PGY_LEVEL], # y=dff['min'], # xaxis=row, # yaxis=col, # fill='tonexty', # fillcolor=lightblue, # line_color=no_color)) # fig.add_trace(go.Scatter(x=dff[PGY_LEVEL], # y=dff['mean'], # xaxis=row, # yaxis=col, # line_color=darkblue, # line_width=4)) # fig.for_each_annotation(lambda a: a.update(text=a.text.split("=")[1])) # fig.update_xaxes(tickvals=[1,2,3,4,5,6], ticktext=['1', '2', '3', 'R', '4', '5']), # fig.update_yaxes(tickvals=[1, 2, 3, 4, 5], ticktext=['OB', 'DS', 'IS', 'UP', 'SO']) # fig.add_hline(y=4, line_dash="dot") # fig.update_layout(title = # 'Operative Comfort By PGY Level', # xaxis_title = 'PGY Level', yaxis_title = 'score', # width = 1200, height = 1600, # font_family='Inter', # showlegend=False, # plot_bgcolor=lightgray) # fig.write_image("op_readiness_curves.svg") # fig.show()

fig = px.line(df_stats, x=PGY_LEVEL, y='min', facet_col=OPERATION, facet_col_wrap=4, facet_row_spacing=0.04, # default is 0.07 when facet_col_wrap is used facet_col_spacing=0.02, # default is 0.03` ) fig.update_traces(line_color=no_color) for t in fig.data: row = t['xaxis'] col = t['yaxis'] proc = t['hovertemplate'].strip('operation=') proc = re.sub('(<br>.+)', '', proc) dff = df_stats.loc[df_stats['operation'] == proc] fig.add_trace(go.Scatter(x=dff[PGY_LEVEL], y=dff['max'], xaxis=row, yaxis=col, fill='tonexty', fillcolor=lightblue, line_color=no_color)) fig.add_trace(go.Scatter(x=dff[PGY_LEVEL], y=dff['min'], xaxis=row, yaxis=col, fill='tonexty', fillcolor=lightblue, line_color=no_color)) fig.add_trace(go.Scatter(x=dff[PGY_LEVEL], y=dff['median'], xaxis=row, yaxis=col, line_color=darkblue, line_width=4)) fig.for_each_annotation(lambda a: a.update(text=a.text.split("=")[1])) fig.update_xaxes(tickvals=['PGY1', 'PGY2', 'PGY3', 'Research', 'PGY4','PGY5'], ticktext=['1', '2', '3', 'R', '4', '5']), fig.update_yaxes(tickvals=[1, 2, 3, 4, 5], ticktext=['OB', 'DS', 'IS', 'UP', 'SO']) fig.add_hline(y=4, line_dash="dot") fig.update_layout(title = 'Operative Comfort By PGY Level', xaxis_title = 'PGY Level', yaxis_title = 'score', width = 1200, height = 1600, font_family='Inter', showlegend=False, plot_bgcolor=lightgray) fig.write_image("op_readiness_curves.svg") fig.show()

# THRESHOLD = 4 df_stats_reverse = df_stats df_stats_reverse df_stats_reverse.replace({PGY_LEVEL: PGY_DOUBLE_REVERESE_DICT}, inplace=True) df_stats_reverse df_threshold = {OPERATION : [], 'threshold_by': []} for name, group in df_stats_reverse.groupby(OPERATION): for index, row in group.iterrows(): if row['median'] >= THRESHOLD: if row[PGY_LEVEL] == 4: level = row[PGY_LEVEL] + 1 else: level = row[PGY_LEVEL] break else: level = 7 df_threshold[OPERATION].append(name) df_threshold['threshold_by'].append(level) df_threshold = pd.DataFrame(df_threshold) df_threshold.sort_values(by=['threshold_by', 'operation'], ascending=False, inplace=True) df_threshold

fig = px.bar(df_threshold, x='threshold_by', y=OPERATION, orientation='h', text_auto=True) fig.update_layout(title = 'Achievement of Unsupervised Practice by Operation', xaxis_title = PGY_LEVEL, yaxis_title = SCORE, width = 1200, height = 800, font_family='Inter', showlegend=False) fig.update_xaxes(tickvals=[1,2,3,4,5,6,7], ticktext=['PGY1', 'PGY2', 'PGY3', 'Research', 'PGY4', 'PGY5', 'Not Reached']) fig.write_image('threshold.svg') fig.update_layout(title = 'PGY Level at Which Unsupervised Practice Threshold is Reached', xaxis_title = 'PGY Level', yaxis_title = 'Activity', width = 1200, height = 1600, font_family='Inter', showlegend=False) fig.show()

SELECT * FROM 'rawdata.csv'