From c54dcf616ab35b7fe84f8795636755d0051db775 Mon Sep 17 00:00:00 2001 From: willemjan Date: Sat, 5 Sep 2026 22:30:10 +0200 Subject: [PATCH 1/2] Add basal/bolus/carb/target overlays to the Live and Dagoverzicht charts MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Overlay the full insulin picture onto the interactive glucose charts: basal rate as a semi-transparent step-area (profile schedule overridden by active temp basal), bolus doses as size-scaled dots split into food/correction/SMB categories (food and correction labeled with units, SMB left unlabeled), carbs as bars, and the pump's target glucose as a line — each with its own color, alpha, and legend entry. Shared trace- building helpers live in ui_common.py so both pages (and the PDF report in a follow-up PR) render identically. Builds on the pump profile work in #29 (profile_data.py, load_profile()). --- pages/1_Live.py | 109 +++++++++++++++++++-------- pages/4_Dagoverzicht.py | 108 ++++++++++++++++++++++++--- reporting.py | 161 ++++++++++++++++++++++++++++++++++++++++ tests/test_reporting.py | 138 ++++++++++++++++++++++++++++++++++ tests/test_ui_common.py | 101 +++++++++++++++++++++++++ ui_common.py | 154 ++++++++++++++++++++++++++++++++++++++ 6 files changed, 729 insertions(+), 42 deletions(-) diff --git a/pages/1_Live.py b/pages/1_Live.py index e653dff..9052a84 100644 --- a/pages/1_Live.py +++ b/pages/1_Live.py @@ -1,6 +1,8 @@ +import pandas as pd import plotly.graph_objects as go import requests import streamlit as st +from plotly.subplots import make_subplots from auth import require_auth from data_processing import ( @@ -10,7 +12,15 @@ from data_processing import ( summarize_time_in_range, ) from nightscout_client import NightscoutConfigError, fetch_entries, fetch_treatments -from ui_common import MMOL_HOVERTEMPLATE, get_colors +from reporting import classify_bolus, effective_basal_series, target_glucose_series +from ui_common import ( + MMOL_HOVERTEMPLATE, + add_basal_area_trace, + add_bolus_marker_traces, + add_target_line_trace, + get_colors, + load_profile, +) st.set_page_config(page_title="Glucose dashboard — Live", page_icon="📊", layout="wide") require_auth() @@ -27,7 +37,28 @@ def load_nightscout_data(hours: int): return glucose, treatments -def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) -> go.Figure: +def _basal_and_target_series(profile: dict, treatments: pd.DataFrame, window_start, window_end): + """Basaal- en streefwaarde-reeksen over een rollend venster dat meerdere + kalenderdagen kan overspannen: per dag berekend (reporting.effective_basal_series/ + target_glucose_series zijn dag-begrensd) en daarna aan elkaar geplakt en + geclipt tot het venster.""" + dates = pd.date_range(window_start.normalize(), window_end.normalize(), freq="D") + basal_parts, target_parts = [], [] + for day_start in dates: + day_end = day_start + pd.Timedelta(days=1) + basal_parts.append(effective_basal_series(profile, treatments, day_start, day_end)) + target_parts.append(target_glucose_series(profile, day_start, day_end)) + + basal = pd.concat(basal_parts, ignore_index=True) if basal_parts else pd.DataFrame(columns=["timestamp", "rate"]) + target = ( + pd.concat(target_parts, ignore_index=True) if target_parts else pd.DataFrame(columns=["timestamp", "target"]) + ) + basal = basal[(basal["timestamp"] >= window_start) & (basal["timestamp"] <= window_end)] + target = target[(target["timestamp"] >= window_start) & (target["timestamp"] <= window_end)] + return basal.reset_index(drop=True), target.reset_index(drop=True) + + +def build_chart(glucose, bolus, carbs, basal, target, site_changes, sensor_changes, colors) -> go.Figure: glucose_min = min(glucose["mmol"].min(), TARGET_LOW) glucose_max = max(glucose["mmol"].max(), TARGET_HIGH) y_top = glucose_max + 1.0 @@ -36,7 +67,7 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - insulin_lane_y = lane_base + 1.5 carbs_lane_y = lane_base + 0.7 - fig = go.Figure() + fig = make_subplots(specs=[[{"secondary_y": True}]]) # Streefbereik als achtergrondband, geen losse legenda-entry nodig. fig.add_hrect( @@ -45,8 +76,13 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - fillcolor=colors["band"], line_width=0, layer="below", + row=1, + col=1, ) + add_basal_area_trace(fig, basal, colors, row=1, secondary_y=True) + add_target_line_trace(fig, target, colors, row=1) + fig.add_trace( go.Scatter( x=glucose["timestamp"], @@ -55,7 +91,9 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - name="Glucose", line={"color": colors["glucose"], "width": 2, "shape": "spline"}, hovertemplate=MMOL_HOVERTEMPLATE, - ) + ), + row=1, + col=1, ) low_points = glucose[glucose["mmol"] < TARGET_LOW] @@ -69,7 +107,9 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - name=f"Laag (<{TARGET_LOW})", marker={"color": colors["low"], "size": 6}, hovertemplate=MMOL_HOVERTEMPLATE, - ) + ), + row=1, + col=1, ) if not high_points.empty: fig.add_trace( @@ -80,21 +120,12 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - name=f"Hoog (>{TARGET_HIGH})", marker={"color": colors["high"], "size": 6}, hovertemplate=MMOL_HOVERTEMPLATE, - ) + ), + row=1, + col=1, ) - if not insulin.empty: - fig.add_trace( - go.Scatter( - x=insulin["timestamp"], - y=[insulin_lane_y] * len(insulin), - mode="markers", - name="Insuline", - marker={"color": colors["insulin"], "size": 10, "symbol": "triangle-up"}, - customdata=insulin["insulin"], - hovertemplate="%{x|%d-%m %H:%M} · %{customdata:.1f} E", - ) - ) + add_bolus_marker_traces(fig, bolus, colors, insulin_lane_y, row=1) if not carbs.empty: fig.add_trace( @@ -106,14 +137,16 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - marker={"color": colors["carbs"], "size": 10, "symbol": "circle"}, customdata=carbs["carbs"], hovertemplate="%{x|%d-%m %H:%M} · %{customdata:.0f} g", - ) + ), + row=1, + col=1, ) # Let op: annotation_text/-position via add_vline() zelf triggert een bug # in plotly's shapeannotation-helper met pandas Timestamp-x-waarden, dus # de lijn en het label worden hier los toegevoegd. for _, row in site_changes.iterrows(): - fig.add_vline(x=row["timestamp"], line_dash="dot", line_color=colors["muted"], opacity=0.6) + fig.add_vline(x=row["timestamp"], line_dash="dot", line_color=colors["muted"], opacity=0.6, row=1, col=1) fig.add_annotation( x=row["timestamp"], y=y_top, @@ -121,9 +154,11 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - text="Pomp", showarrow=False, font={"color": colors["muted"], "size": 10}, + row=1, + col=1, ) for _, row in sensor_changes.iterrows(): - fig.add_vline(x=row["timestamp"], line_dash="dash", line_color=colors["muted"], opacity=0.6) + fig.add_vline(x=row["timestamp"], line_dash="dash", line_color=colors["muted"], opacity=0.6, row=1, col=1) fig.add_annotation( x=row["timestamp"], y=y_top, @@ -131,6 +166,8 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - text="Sensor", showarrow=False, font={"color": colors["muted"], "size": 10}, + row=1, + col=1, ) fig.update_layout( @@ -140,14 +177,18 @@ def build_chart(glucose, insulin, carbs, site_changes, sensor_changes, colors) - legend={"orientation": "h", "yanchor": "bottom", "y": 1.02, "xanchor": "left", "x": 0}, margin={"l": 10, "r": 10, "t": 10, "b": 10}, height=480, - yaxis={ - "title": "mmol/L", - "range": [y_bottom, y_top], - "gridcolor": colors["grid"], - "zeroline": False, - }, - xaxis={"gridcolor": colors["grid"]}, ) + fig.update_yaxes( + title_text="mmol/L", + range=[y_bottom, y_top], + gridcolor=colors["grid"], + zeroline=False, + row=1, + col=1, + secondary_y=False, + ) + fig.update_yaxes(title_text="Basaal (E/uur)", showgrid=False, row=1, col=1, secondary_y=True) + fig.update_xaxes(gridcolor=colors["grid"], row=1, col=1) return fig @@ -160,6 +201,7 @@ with st.sidebar: try: glucose, treatments = load_nightscout_data(hours) + profile = load_profile() except NightscoutConfigError as exc: st.error(str(exc)) st.stop() @@ -172,7 +214,9 @@ if glucose.empty: st.stop() events = classify_treatments(treatments) +bolus = classify_bolus(treatments) tir = summarize_time_in_range(glucose) +basal, target = _basal_and_target_series(profile, treatments, glucose["timestamp"].min(), glucose["timestamp"].max()) latest = glucose.iloc[-1] previous = glucose.iloc[-2] if len(glucose) > 1 else None @@ -192,14 +236,17 @@ colors = get_colors() fig = build_chart( glucose, - events["insulin"], + bolus, events["carbs"], + basal, + target, events["site_changes"], events["sensor_changes"], colors, ) st.plotly_chart(fig, width="stretch") st.caption( - "Driehoekjes = insuline · cirkels = koolhydraten · stippellijn = pompwissel · " - "streepjeslijn = sensorwissel" + "Stippen = bolusinsuline (voeding/correctie/SMB, grootte naar dosis) · cirkels = koolhydraten · " + "vlak = basaal · streepstippellijn = streefwaarde uit pompprofiel · " + "stippellijn = pompwissel · streepjeslijn = sensorwissel" ) diff --git a/pages/4_Dagoverzicht.py b/pages/4_Dagoverzicht.py index 110e278..6584a08 100644 --- a/pages/4_Dagoverzicht.py +++ b/pages/4_Dagoverzicht.py @@ -1,14 +1,21 @@ +import pandas as pd import plotly.graph_objects as go import requests import streamlit as st from plotly.subplots import make_subplots from auth import require_auth -from data_processing import TARGET_HIGH, TARGET_LOW +from data_processing import TARGET_HIGH, TARGET_LOW, classify_treatments from nightscout_client import NightscoutConfigError +from reporting import classify_bolus, effective_basal_series, target_glucose_series from ui_common import ( + add_basal_area_trace, + add_bolus_marker_traces, + add_carb_bars_trace, + add_target_line_trace, add_threshold_lines, get_colors, + load_profile, load_range_data, render_completeness_warning, select_period_days, @@ -18,7 +25,7 @@ st.set_page_config(page_title="Glucose dashboard — Dagoverzicht", page_icon=" require_auth() st.title("Dagoverzicht") -st.caption("Eén glucosegrafiek per kalenderdag (FR4).") +st.caption("Eén glucosegrafiek per kalenderdag, met basaal/bolus/koolhydraten-overlay (FR4).") start, end, days = select_period_days(default_label="14 dagen") if days > 31: @@ -28,7 +35,8 @@ if days > 31: ) try: - glucose, _treatments = load_range_data(start, end) + glucose, treatments = load_range_data(start, end) + profile = load_profile() except NightscoutConfigError as exc: st.error(str(exc)) st.stop() @@ -47,44 +55,122 @@ glucose = glucose.copy() glucose["date"] = glucose["timestamp"].dt.date glucose["hour"] = glucose["timestamp"].dt.hour + glucose["timestamp"].dt.minute / 60 +carbs_events = classify_treatments(treatments)["carbs"].copy() +if not carbs_events.empty: + carbs_events["date"] = carbs_events["timestamp"].dt.date + carbs_events["hour"] = carbs_events["timestamp"].dt.hour + carbs_events["timestamp"].dt.minute / 60 + +bolus_events = classify_bolus(treatments) +if not bolus_events.empty: + bolus_events["date"] = bolus_events["timestamp"].dt.date + bolus_events["hour"] = bolus_events["timestamp"].dt.hour + bolus_events["timestamp"].dt.minute / 60 + + +def _close_step_to_day_end(df: pd.DataFrame, value_col: str) -> pd.DataFrame: + """Voeg een laatste punt op hour=24 toe (zelfde waarde als het laatste + segment) zodat een stapvlak/-lijn de volle dag doortekent i.p.v. te + stoppen bij het laatste breekpunt.""" + if df.empty: + return df + last_value = df.iloc[-1][value_col] + closing_row = pd.DataFrame([{"hour": 24.0, value_col: last_value}]) + return pd.concat([df, closing_row], ignore_index=True) + + dates = sorted(glucose["date"].unique()) -row_height = 150 +row_height = 190 fig = make_subplots( rows=len(dates), cols=1, shared_xaxes=True, - subplot_titles=[d.strftime("%A %d-%m-%Y") for d in dates], vertical_spacing=min(0.4 / max(len(dates) - 1, 1), 0.05), + specs=[[{"secondary_y": True}]] * len(dates), ) for i, day in enumerate(dates, start=1): day_df = glucose[glucose["date"] == day] + day_start = pd.Timestamp(day) + day_end = day_start + pd.Timedelta(days=1) + fig.add_hrect( y0=TARGET_LOW, y1=TARGET_HIGH, fillcolor=colors["band"], line_width=0, layer="below", row=i, col=1 ) + # Handmatige titel-annotatie t.o.v. het domein van déze rij i.p.v. + # make_subplots' subplot_titles (die t.o.v. de hele figuur positioneert en + # daardoor bij veel dagen de gedeelde legenda bovenaan kan overlappen). + fig.add_annotation( + text=day.strftime("%A %d-%m-%Y"), + xref="x domain", + yref="y domain", + x=0.0, + y=1.15, + xanchor="left", + yanchor="bottom", + showarrow=False, + font={"size": 13, "color": colors["text"]}, + row=i, + col=1, + ) + + basal_day = effective_basal_series(profile, treatments, day_start, day_end) + if not basal_day.empty: + basal_day["hour"] = (basal_day["timestamp"] - day_start).dt.total_seconds() / 3600 + basal_day = _close_step_to_day_end(basal_day, "rate") + add_basal_area_trace(fig, basal_day, colors, row=i, secondary_y=True, x_col="hour") + + target_day = target_glucose_series(profile, day_start, day_end) + if not target_day.empty: + target_day["hour"] = (target_day["timestamp"] - day_start).dt.total_seconds() / 3600 + target_day = _close_step_to_day_end(target_day, "target") + add_target_line_trace(fig, target_day, colors, row=i, x_col="hour") + + add_carb_bars_trace(fig, carbs_events[carbs_events["date"] == day], colors, row=i, secondary_y=True, x_col="hour") + + lane_y = day_df["mmol"].min() - 1.0 if not day_df.empty else TARGET_LOW - 1.0 + add_bolus_marker_traces(fig, bolus_events[bolus_events["date"] == day], colors, lane_y, row=i, x_col="hour") + fig.add_trace( go.Scatter( x=day_df["hour"], y=day_df["mmol"], mode="lines", line={"color": colors["glucose"], "width": 2}, - showlegend=False, + name="Glucose", hovertemplate="%{y:.1f} mmol/L", ), row=i, col=1, + secondary_y=False, + ) + fig.update_yaxes(title_text="mmol/L", gridcolor=colors["grid"], row=i, col=1, secondary_y=False) + fig.update_yaxes( + title_text="Basaal (E/uur) · KH (g)", showgrid=False, row=i, col=1, secondary_y=True ) - fig.update_yaxes(title_text="mmol/L", gridcolor=colors["grid"], row=i, col=1) add_threshold_lines(fig, colors, row=i) +# Elke dag herhaalt dezelfde spoornamen (Glucose/Basaal/Koolhydraten/...) — zonder +# deduplicatie zou de legenda 7x "Glucose" etc. tonen. Alleen de eerste keer per +# naam blijft in de legenda staan. +seen_names = set() +for trace in fig.data: + if trace.name in seen_names: + trace.showlegend = False + else: + seen_names.add(trace.name) + fig.update_xaxes(title_text="Uur van de dag", range=[0, 24], dtick=4, gridcolor=colors["grid"]) fig.update_layout( paper_bgcolor="rgba(0,0,0,0)", plot_bgcolor="rgba(0,0,0,0)", font_color=colors["text"], - height=row_height * len(dates), - showlegend=False, - margin={"l": 10, "r": 10, "t": 30, "b": 10}, + height=row_height * len(dates) + 40, + showlegend=True, + legend={"orientation": "h", "yanchor": "bottom", "y": 1.0, "xanchor": "left", "x": 0}, + margin={"l": 10, "r": 10, "t": 40, "b": 10}, ) st.plotly_chart(fig, width="stretch") -st.caption(f"{len(dates)} dagen · {len(glucose)} metingen") +st.caption( + f"{len(dates)} dagen · {len(glucose)} metingen · stippen = bolusinsuline " + "(voeding/correctie/SMB, grootte naar dosis) · vlak = basaal · staafjes = koolhydraten · " + "streepstippellijn = streefwaarde uit pompprofiel" +) diff --git a/reporting.py b/reporting.py index d59cf58..1c881d7 100644 --- a/reporting.py +++ b/reporting.py @@ -15,6 +15,7 @@ import numpy as np import pandas as pd from nightscout_client import INSULIN_EVENT_TYPES +from profile_data import value_at # Klinische drempels in mmol/L (Battelino et al. 2019, Diabetes Care 42(8):1593-1603). VERY_LOW = 3.0 @@ -440,6 +441,166 @@ def bolus_stats(treatments: pd.DataFrame) -> dict: } +BOLUS_CATEGORY_FOOD = "food" +BOLUS_CATEGORY_CORRECTION = "correction" +BOLUS_CATEGORY_SMB = "smb" + +BOLUS_CATEGORY_COLUMNS = ["timestamp", "eventType", "insulin", "carbs", "category"] + + +def classify_bolus(treatments: pd.DataFrame) -> pd.DataFrame: + """Splits bolus-insuline op in 'food'/'correction'/'smb', voor grafiekoverlays + (gitea insulineprofiel-werk, vervolg op bolus_stats()'s SMB-detectie). + + Herkenningsregels, in volgorde: + - SMB: `eventType == "SMB"`, of `eventType == "Correction Bolus"` met + `type == "SMB"` (zie bolus_stats()) — automatische AndroidAPS-microbolus. + - food: `eventType == "Meal Bolus"`, of een andere bolus-eventType + (`Bolus`/`Combo Bolus`) met koolhydraten op dezelfde treatment-rij (een + sterke aanwijzing dat het om een maaltijdbolus gaat). + - correction: alle overige bolus-rijen (typisch `Correction Bolus` zonder + SMB-markering, of een generieke `Bolus`/`Combo Bolus` zonder koolhydraten). + + Geeft een DataFrame terug met kolommen 'timestamp', 'eventType', 'insulin', + 'carbs', 'category' — alleen de rijen die daadwerkelijk bolusinsuline zijn. + """ + if treatments.empty: + return pd.DataFrame(columns=BOLUS_CATEGORY_COLUMNS) + + bolus = treatments[treatments["eventType"].isin(INSULIN_EVENT_TYPES) & treatments["insulin"].notna()].copy() + if bolus.empty: + return pd.DataFrame(columns=BOLUS_CATEGORY_COLUMNS) + + is_smb = (bolus["eventType"] == SMB_TYPE) | ( + (bolus["eventType"] == "Correction Bolus") & (bolus.get("type", "") == SMB_TYPE) + ) + has_carbs = pd.to_numeric(bolus["carbs"], errors="coerce").fillna(0) > 0 + is_food = ~is_smb & ((bolus["eventType"] == "Meal Bolus") | has_carbs) + + bolus["category"] = BOLUS_CATEGORY_CORRECTION + bolus.loc[is_food, "category"] = BOLUS_CATEGORY_FOOD + bolus.loc[is_smb, "category"] = BOLUS_CATEGORY_SMB + return bolus[BOLUS_CATEGORY_COLUMNS].sort_values("timestamp").reset_index(drop=True) + + +def glucose_min_max(glucose: pd.DataFrame) -> tuple[float, float]: + """Minimum en maximum glucosewaarde (mmol/L) in de gegeven data.""" + if glucose.empty: + return (0.0, 0.0) + return (round(glucose["mmol"].min(), 1), round(glucose["mmol"].max(), 1)) + + +def _temp_basal_intervals(treatments: pd.DataFrame, day_start, day_end) -> list[tuple]: + """Temp Basal-intervallen (start, eind, rate) die (deels) binnen [day_start, day_end) vallen, + geclipt aan de daggrenzen.""" + if treatments is None or treatments.empty: + return [] + temp_basal = treatments[ + (treatments["eventType"] == BASAL_EVENT_TYPE) + & treatments["absolute"].notna() + & treatments["duration"].notna() + ] + intervals = [] + for _, row in temp_basal.iterrows(): + start = row["timestamp"] + end = start + pd.Timedelta(float(row["duration"]), unit="min") + if end <= day_start or start >= day_end: + continue + intervals.append((max(start, day_start), min(end, day_end), row["absolute"])) + return intervals + + +def _basal_breakpoints(basal_segments: list[dict], temp_intervals: list[tuple], day_start, day_end) -> list: + """Alle tijdstippen waarop de effectieve basaalsnelheid kan wijzigen: daggrens, + profielsegment-grenzen, en temp-basal start/eind — oplopend gesorteerd.""" + breakpoints = {day_start} + for seg in basal_segments: + ts = day_start + pd.Timedelta(float(seg["minutes"]), unit="min") + if day_start <= ts < day_end: + breakpoints.add(ts) + for start, end, _ in temp_intervals: + breakpoints.add(start) + breakpoints.add(end) + return sorted(t for t in breakpoints if day_start <= t < day_end) + + +def _rate_at(ts, basal_segments: list[dict], temp_intervals: list[tuple], day_start) -> float: + """Effectieve basaalsnelheid op tijdstip `ts`: een actieve Temp Basal wint van het profieltempo.""" + for start, end, rate in temp_intervals: + if start <= ts < end: + return rate + minutes_of_day = int((ts - day_start).total_seconds() / 60) % MINUTES_PER_DAY + value = value_at(basal_segments, minutes_of_day) + return value if value is not None else 0.0 + + +def effective_basal_series(profile: dict | None, treatments: pd.DataFrame, day_start, day_end) -> pd.DataFrame: + """Basaalsnelheid (E/uur) als stapfunctie over [day_start, day_end). + + Basislijn is het pompprofiel se basaalschema (`profile["basal"]`); waar een + Temp Basal-treatment actief is (rate/duur uit de treatment) overschrijft + die het profieltempo voor de duur van die tijdelijke aanpassing — dit is de + daadwerkelijk afgeleverde basaal, in tegenstelling tot basal_stats() (die + alleen temp-basal-periodes optelt en periodes zonder actieve temp basal + mist omdat het profiel daar niet bij betrokken wordt). + + Zonder profiel (`profile` is None/leeg) is de basislijn 0 — dan toont dit + alleen de temp-basal-periodes, net als basal_stats(). + + Geeft een DataFrame terug met kolommen 'timestamp' en 'rate': elke rij + geldt tot de volgende rij (stapfunctie), geschikt voor een + step-area-plot (line_shape='hv', fill='tozeroy'). + """ + day_start = pd.Timestamp(day_start) + day_end = pd.Timestamp(day_end) + basal_segments = (profile or {}).get("basal", []) + temp_intervals = _temp_basal_intervals(treatments, day_start, day_end) + breakpoints = _basal_breakpoints(basal_segments, temp_intervals, day_start, day_end) + + rows = [ + {"timestamp": ts, "rate": _rate_at(ts, basal_segments, temp_intervals, day_start)} for ts in breakpoints + ] + return pd.DataFrame(rows, columns=["timestamp", "rate"]) + + +def _target_midpoint_at(minutes_of_day: int, target_low: list[dict], target_high: list[dict]) -> float | None: + low = value_at(target_low, minutes_of_day) if target_low else None + high = value_at(target_high, minutes_of_day) if target_high else None + if low is not None and high is not None: + return (low + high) / 2 + return low if low is not None else high + + +def target_glucose_series(profile: dict | None, day_start, day_end) -> pd.DataFrame: + """Streefglucose (mmol/L) uit het profiel als stapfunctie over [day_start, day_end) — + het midden van target_low/target_high per tijdsegment (de pomp kent vaak één + streefwaarde per segment; met losse low/high-segmenten is dat gewoon het + gemiddelde). Geeft een lege DataFrame terug als het profiel geen streefwaarden + bevat. + + Zelfde stapfunctie-vorm als effective_basal_series(): elke rij geldt tot de + volgende rij. + """ + day_start = pd.Timestamp(day_start) + day_end = pd.Timestamp(day_end) + target_low = (profile or {}).get("target_low", []) + target_high = (profile or {}).get("target_high", []) + if not target_low and not target_high: + return pd.DataFrame(columns=["timestamp", "target"]) + + breakpoints = _basal_breakpoints(target_low + target_high, [], day_start, day_end) + rows = [ + { + "timestamp": ts, + "target": _target_midpoint_at( + int((ts - day_start).total_seconds() / 60) % MINUTES_PER_DAY, target_low, target_high + ), + } + for ts in breakpoints + ] + return pd.DataFrame(rows, columns=["timestamp", "target"]) + + def basal_stats(treatments: pd.DataFrame, period_days: float) -> dict: """Totale basaalinsuline uit Temp Basal-treatments (rate x duur), zonder pompprofiel. diff --git a/tests/test_reporting.py b/tests/test_reporting.py index b981d83..5eca552 100644 --- a/tests/test_reporting.py +++ b/tests/test_reporting.py @@ -8,10 +8,13 @@ from reporting import ( agp_percentile_bands, basal_stats, bolus_stats, + classify_bolus, completeness_warning, daily_gmi, day_of_week_pattern, detect_events, + effective_basal_series, + glucose_min_max, gmi_mmol_per_mol, gmi_percent, gmi_trend, @@ -21,6 +24,7 @@ from reporting import ( sensor_active_percent, site_interval_stats, summarize_events, + target_glucose_series, tir_breakdown_5band, ) @@ -480,6 +484,140 @@ def test_basal_stats_handles_empty_input(): assert result == {"total_units": 0.0, "avg_units_per_day": 0.0} +def test_glucose_min_max_returns_min_and_max(): + glucose = _series([("2026-01-01 00:00", 6.5), ("2026-01-01 01:00", 3.2), ("2026-01-01 02:00", 14.1)]) + assert glucose_min_max(glucose) == (3.2, 14.1) + + +def test_glucose_min_max_handles_empty_input(): + empty = pd.DataFrame(columns=["timestamp", "mmol"]) + assert glucose_min_max(empty) == (0.0, 0.0) + + +def _bolus_treatment(dt, event_type, insulin=1.0, carbs=None, bolus_type=""): + return { + "timestamp": pd.Timestamp(dt), + "eventType": event_type, + "insulin": insulin, + "carbs": carbs, + "type": bolus_type, + } + + +def test_classify_bolus_recognizes_meal_bolus_as_food(): + treatments = pd.DataFrame([_bolus_treatment("2026-01-01 08:00", "Meal Bolus", insulin=4.0)]) + result = classify_bolus(treatments) + assert result.iloc[0]["category"] == "food" + + +def test_classify_bolus_recognizes_smb_via_type_field(): + treatments = pd.DataFrame( + [_bolus_treatment("2026-01-01 08:00", "Correction Bolus", insulin=0.3, bolus_type="SMB")] + ) + result = classify_bolus(treatments) + assert result.iloc[0]["category"] == "smb" + + +def test_classify_bolus_recognizes_smb_via_event_type(): + treatments = pd.DataFrame([_bolus_treatment("2026-01-01 08:00", "SMB", insulin=0.3)]) + result = classify_bolus(treatments) + assert result.iloc[0]["category"] == "smb" + + +def test_classify_bolus_treats_correction_bolus_without_carbs_as_correction(): + treatments = pd.DataFrame([_bolus_treatment("2026-01-01 08:00", "Correction Bolus", insulin=1.5)]) + result = classify_bolus(treatments) + assert result.iloc[0]["category"] == "correction" + + +def test_classify_bolus_treats_generic_bolus_with_carbs_as_food(): + treatments = pd.DataFrame([_bolus_treatment("2026-01-01 08:00", "Bolus", insulin=3.0, carbs=40.0)]) + result = classify_bolus(treatments) + assert result.iloc[0]["category"] == "food" + + +def test_classify_bolus_ignores_non_bolus_rows(): + treatments = pd.DataFrame( + [ + _bolus_treatment("2026-01-01 08:00", "Meal Bolus", insulin=4.0), + {"timestamp": pd.Timestamp("2026-01-01 08:05"), "eventType": "Site Change", "insulin": None, "carbs": None, "type": ""}, + ] + ) + result = classify_bolus(treatments) + assert len(result) == 1 + + +def test_classify_bolus_handles_empty_input(): + empty = pd.DataFrame(columns=["timestamp", "eventType", "insulin", "carbs", "type"]) + result = classify_bolus(empty) + assert result.empty + assert list(result.columns) == ["timestamp", "eventType", "insulin", "carbs", "category"] + + +def test_effective_basal_series_uses_profile_baseline_without_temp_basal(): + profile = {"basal": [{"minutes": 0, "value": 0.8}, {"minutes": 360, "value": 1.2}]} + empty_treatments = pd.DataFrame(columns=["timestamp", "eventType", "absolute", "duration"]) + result = effective_basal_series( + profile, empty_treatments, pd.Timestamp("2026-01-01"), pd.Timestamp("2026-01-02") + ) + assert result.iloc[0]["rate"] == pytest.approx(0.8) + assert result[result["timestamp"] == pd.Timestamp("2026-01-01 06:00")].iloc[0]["rate"] == pytest.approx(1.2) + + +def test_effective_basal_series_temp_basal_overrides_profile_rate(): + profile = {"basal": [{"minutes": 0, "value": 0.8}]} + treatments = pd.DataFrame( + [{"timestamp": pd.Timestamp("2026-01-01 08:00"), "eventType": "Temp Basal", "absolute": 2.0, "duration": 30}] + ) + result = effective_basal_series(profile, treatments, pd.Timestamp("2026-01-01"), pd.Timestamp("2026-01-02")) + during_temp = result[result["timestamp"] == pd.Timestamp("2026-01-01 08:00")].iloc[0] + after_temp = result[result["timestamp"] == pd.Timestamp("2026-01-01 08:30")].iloc[0] + assert during_temp["rate"] == pytest.approx(2.0) + assert after_temp["rate"] == pytest.approx(0.8) # terug naar het profieltempo + + +def test_effective_basal_series_without_profile_defaults_to_zero_baseline(): + empty_treatments = pd.DataFrame(columns=["timestamp", "eventType", "absolute", "duration"]) + result = effective_basal_series(None, empty_treatments, pd.Timestamp("2026-01-01"), pd.Timestamp("2026-01-02")) + assert result.iloc[0]["rate"] == pytest.approx(0.0) + + +def test_effective_basal_series_clips_temp_basal_to_day_boundaries(): + profile = {"basal": [{"minutes": 0, "value": 0.8}]} + # Temp basal begint de dag ervoor en loopt door tot 01:00 van de dag zelf. + treatments = pd.DataFrame( + [{"timestamp": pd.Timestamp("2025-12-31 23:00"), "eventType": "Temp Basal", "absolute": 2.0, "duration": 120}] + ) + result = effective_basal_series(profile, treatments, pd.Timestamp("2026-01-01"), pd.Timestamp("2026-01-02")) + assert result.iloc[0]["timestamp"] == pd.Timestamp("2026-01-01") + assert result.iloc[0]["rate"] == pytest.approx(2.0) + + +def test_target_glucose_series_averages_low_and_high(): + profile = { + "target_low": [{"minutes": 0, "value": 4.5}], + "target_high": [{"minutes": 0, "value": 8.5}], + } + result = target_glucose_series(profile, pd.Timestamp("2026-01-01"), pd.Timestamp("2026-01-02")) + assert result.iloc[0]["target"] == pytest.approx(6.5) + + +def test_target_glucose_series_falls_back_to_single_bound(): + profile = {"target_low": [{"minutes": 0, "value": 5.0}], "target_high": []} + result = target_glucose_series(profile, pd.Timestamp("2026-01-01"), pd.Timestamp("2026-01-02")) + assert result.iloc[0]["target"] == pytest.approx(5.0) + + +def test_target_glucose_series_empty_without_targets_in_profile(): + result = target_glucose_series({}, pd.Timestamp("2026-01-01"), pd.Timestamp("2026-01-02")) + assert result.empty + + +def test_target_glucose_series_handles_none_profile(): + result = target_glucose_series(None, pd.Timestamp("2026-01-01"), pd.Timestamp("2026-01-02")) + assert result.empty + + def test_site_interval_stats_splits_by_site_change(): glucose = _series( [ diff --git a/tests/test_ui_common.py b/tests/test_ui_common.py index 38ab716..c2829e6 100644 --- a/tests/test_ui_common.py +++ b/tests/test_ui_common.py @@ -1,8 +1,13 @@ +import pandas as pd import plotly.graph_objects as go from ui_common import ( PALETTE, THRESHOLD_LINES, + add_basal_area_trace, + add_bolus_marker_traces, + add_carb_bars_trace, + add_target_line_trace, add_threshold_lines, build_tir_stacked_bar_figure, hex_to_rgba, @@ -36,3 +41,99 @@ def test_build_tir_stacked_bar_figure_has_one_bar_trace_per_band(): assert len(fig.data) == 5 assert [trace.y[0] for trace in fig.data] == [1.0, 3.0, 80.0, 12.0, 4.0] assert fig.layout.barmode == "stack" + + +def test_add_basal_area_trace_adds_one_filled_step_trace(): + basal = pd.DataFrame( + {"timestamp": pd.to_datetime(["2026-01-01 00:00", "2026-01-01 06:00"]), "rate": [0.8, 1.2]} + ) + fig = go.Figure() + add_basal_area_trace(fig, basal, PALETTE["light"]) + assert len(fig.data) == 1 + assert fig.data[0].name == "Basaal" + assert fig.data[0].fill == "tozeroy" + assert fig.data[0].line.shape == "hv" + + +def test_add_basal_area_trace_skips_empty_input(): + fig = go.Figure() + add_basal_area_trace(fig, pd.DataFrame(columns=["timestamp", "rate"]), PALETTE["light"]) + assert len(fig.data) == 0 + + +def test_add_carb_bars_trace_adds_one_bar_trace(): + carbs = pd.DataFrame({"timestamp": pd.to_datetime(["2026-01-01 08:00"]), "carbs": [40.0]}) + fig = go.Figure() + add_carb_bars_trace(fig, carbs, PALETTE["light"]) + assert len(fig.data) == 1 + assert fig.data[0].name == "Koolhydraten" + assert isinstance(fig.data[0], go.Bar) + + +def test_add_bolus_marker_traces_adds_one_trace_per_category_present(): + bolus = pd.DataFrame( + { + "timestamp": pd.to_datetime(["2026-01-01 08:00", "2026-01-01 12:00"]), + "insulin": [4.0, 0.3], + "category": ["food", "smb"], + } + ) + fig = go.Figure() + add_bolus_marker_traces(fig, bolus, PALETTE["light"], lane_y=2.0) + names = {trace.name for trace in fig.data} + assert names == {"Bolus (voeding)", "SMB"} + + +def test_add_bolus_marker_traces_labels_food_and_correction_but_not_smb(): + bolus = pd.DataFrame( + { + "timestamp": pd.to_datetime(["2026-01-01 08:00", "2026-01-01 12:00"]), + "insulin": [4.0, 0.3], + "category": ["food", "smb"], + } + ) + fig = go.Figure() + add_bolus_marker_traces(fig, bolus, PALETTE["light"], lane_y=2.0) + by_name = {trace.name: trace for trace in fig.data} + assert by_name["Bolus (voeding)"].mode == "markers+text" + assert by_name["SMB"].mode == "markers" + + +def test_add_bolus_marker_traces_skips_empty_input(): + fig = go.Figure() + add_bolus_marker_traces(fig, pd.DataFrame(columns=["timestamp", "insulin", "category"]), PALETTE["light"], 2.0) + assert len(fig.data) == 0 + + +def test_add_target_line_trace_adds_one_line_trace(): + target = pd.DataFrame({"timestamp": pd.to_datetime(["2026-01-01 00:00"]), "target": [6.5]}) + fig = go.Figure() + add_target_line_trace(fig, target, PALETTE["light"]) + assert len(fig.data) == 1 + assert fig.data[0].name == "Streefwaarde (pomp)" + + +def test_add_target_line_trace_skips_empty_input(): + fig = go.Figure() + add_target_line_trace(fig, pd.DataFrame(columns=["timestamp", "target"]), PALETTE["light"]) + assert len(fig.data) == 0 + + +def test_overlay_traces_support_an_hour_based_x_column_for_dagoverzicht(): + """Dagoverzicht plot elke dag op een gedeelde 0-24u-as ('hour') i.p.v. echte + datums; de hover mag daar dan ook geen datumformattering tonen.""" + basal = pd.DataFrame({"hour": [0.0, 6.0], "rate": [0.8, 1.2]}) + carbs = pd.DataFrame({"hour": [8.0], "carbs": [40.0]}) + bolus = pd.DataFrame({"hour": [8.0], "insulin": [4.0], "category": ["food"]}) + target = pd.DataFrame({"hour": [0.0], "target": [6.5]}) + + fig = go.Figure() + add_basal_area_trace(fig, basal, PALETTE["light"], x_col="hour") + add_carb_bars_trace(fig, carbs, PALETTE["light"], x_col="hour") + add_bolus_marker_traces(fig, bolus, PALETTE["light"], lane_y=2.0, x_col="hour") + add_target_line_trace(fig, target, PALETTE["light"], x_col="hour") + + assert len(fig.data) == 4 + bolus_trace = next(t for t in fig.data if t.name == "Bolus (voeding)") + assert list(bolus_trace.x) == [8.0] + assert "%d-%m" not in bolus_trace.hovertemplate diff --git a/ui_common.py b/ui_common.py index bd81d6f..554e74a 100644 --- a/ui_common.py +++ b/ui_common.py @@ -209,6 +209,160 @@ def build_tir_stacked_bar_figure(tir: dict, colors: dict) -> go.Figure: return fig +BOLUS_CATEGORY_LABELS = { + "food": "Bolus (voeding)", + "correction": "Bolus (correctie)", + "smb": "SMB", +} +BOLUS_CATEGORY_COLOR_KEYS = { + "food": "bolus_food", + "correction": "bolus_correction", + "smb": "bolus_smb", +} +# Tekenvolgorde van bolus-categorieën: SMB eerst (onderop, meestal talrijker en +# kleiner), voeding/correctie erbovenop (grotere, gelabelde stippen blijven +# zichtbaar in plaats van bedolven te raken onder de SMB-stippen). +BOLUS_CATEGORY_DRAW_ORDER = ["smb", "correction", "food"] + + +def add_basal_area_trace( + fig: go.Figure, + basal: pd.DataFrame, + colors: dict, + *, + row: int | None = None, + secondary_y: bool = False, + x_col: str = "timestamp", +) -> None: + """Basaal-insuline als semi-transparant stapvlak (line_shape='hv'), met alpha + zodat onderliggende data zichtbaar blijft. Verwacht kolommen 'rate' en + `x_col` (standaard 'timestamp'; de Dagoverzicht-pagina geeft hier 'hour' + aan omdat elke dag daar op een gedeelde 0-24u-as staat i.p.v. echte + datums — zie reporting.effective_basal_series). Bedoeld voor een + secundaire y-as (E/uur past niet op de mmol/L-schaal van de glucoselijn). + """ + if basal.empty: + return + kwargs: dict = {"row": row, "col": 1} if row is not None else {} + if secondary_y: + kwargs["secondary_y"] = True + fig.add_trace( + go.Scatter( + x=basal[x_col], + y=basal["rate"], + mode="lines", + line={"shape": "hv", "color": colors["basal"], "width": 1}, + fill="tozeroy", + fillcolor=hex_to_rgba(colors["basal"], 0.3), + opacity=0.8, + name="Basaal", + hovertemplate="%{y:.2f} E/uur", + ), + **kwargs, + ) + + +def add_carb_bars_trace( + fig: go.Figure, + carbs: pd.DataFrame, + colors: dict, + *, + row: int | None = None, + secondary_y: bool = False, + x_col: str = "timestamp", +) -> None: + """Koolhydraten als staafjes op het moment van invoer. Verwacht kolommen + 'carbs' en `x_col` (zie add_basal_area_trace voor waarom dit configureerbaar + is). Zelfde secundaire as als de basaal-area (allebei geen mmol/L-grootheid), + met een eigen kleur en alpha zodat beide onderscheidbaar blijven.""" + if carbs.empty: + return + kwargs: dict = {"row": row, "col": 1} if row is not None else {} + if secondary_y: + kwargs["secondary_y"] = True + fig.add_trace( + go.Bar( + x=carbs[x_col], + y=carbs["carbs"], + marker_color=colors["carbs"], + opacity=0.55, + name="Koolhydraten", + hovertemplate="%{y:.0f} g", + ), + **kwargs, + ) + + +def add_bolus_marker_traces( + fig: go.Figure, + bolus: pd.DataFrame, + colors: dict, + lane_y: float, + *, + row: int | None = None, + x_col: str = "timestamp", +) -> None: + """Eén Scatter-trace per boluscategorie (food/correction/smb), gepositioneerd + op een vaste 'lane'-hoogte op de primaire (mmol/L-)as — net als de bestaande + insuline/carbs-lanes op de Live-pagina. De hoogte zelf draagt geen + kwantitatieve betekenis; de markergrootte (geschaald naar toegediende + eenheden) en hoverwaarde wel. Voeding/correctie krijgen een tekstlabel met + de dosis; SMB's niet (te talrijk/te klein voor leesbare labels). Verwacht + kolommen 'insulin'/'category' en `x_col` (zie reporting.classify_bolus). + """ + # Datumformattering in de hover werkt alleen op een echte datetime-as; de + # Dagoverzicht-pagina geeft hier 'hour' (een kaal getal 0-24) aan, dus + # daar laten we het tijdstip weg en toont de hover alleen de dosis. + hover_prefix = "%{x|%d-%m %H:%M} · " if x_col == "timestamp" else "" + kwargs: dict = {"row": row, "col": 1} if row is not None else {} + for category in BOLUS_CATEGORY_DRAW_ORDER: + subset = bolus[bolus["category"] == category] if not bolus.empty else bolus + if subset.empty: + continue + show_label = category != "smb" + fig.add_trace( + go.Scatter( + x=subset[x_col], + y=[lane_y] * len(subset), + mode="markers+text" if show_label else "markers", + marker={ + "size": (subset["insulin"] * 8).clip(lower=6, upper=30), + "color": colors[BOLUS_CATEGORY_COLOR_KEYS[category]], + "opacity": 0.75, + }, + text=[f"{v:.1f}E" for v in subset["insulin"]] if show_label else None, + textposition="top center", + textfont={"size": 9, "color": colors["text"]}, + name=BOLUS_CATEGORY_LABELS[category], + customdata=subset["insulin"], + hovertemplate=f"{hover_prefix}%{{customdata:.2f}} E", + ), + **kwargs, + ) + + +def add_target_line_trace( + fig: go.Figure, target: pd.DataFrame, colors: dict, *, row: int | None = None, x_col: str = "timestamp" +) -> None: + """Streefglucose uit het pompprofiel als lijn op de primaire (mmol/L-)as. + Verwacht kolommen 'target' en `x_col` (zie reporting.target_glucose_series).""" + if target.empty: + return + kwargs: dict = {"row": row, "col": 1} if row is not None else {} + fig.add_trace( + go.Scatter( + x=target[x_col], + y=target["target"], + mode="lines", + line={"shape": "hv", "color": colors["target"], "width": 1.5, "dash": "dashdot"}, + opacity=0.9, + name="Streefwaarde (pomp)", + hovertemplate="%{y:.1f} mmol/L", + ), + **kwargs, + ) + + def render_completeness_warning(glucose: pd.DataFrame, start: date, end: date) -> None: """Toon reporting.completeness_warning()'s oordeel als st.warning, indien nodig. -- 2.54.0 From d33516e2a10ee292df228fc4b8683fa5f2850e59 Mon Sep 17 00:00:00 2001 From: willemjan Date: Sat, 5 Sep 2026 22:38:07 +0200 Subject: [PATCH 2/2] Avoid pandas Timedelta(days=1) deprecation warning pd.Timedelta(days=1) triggers a DeprecationWarning about NumPy's "generic" timedelta unit in this pandas version; pd.Timedelta(1, unit="D") is equivalent and warning-free. --- pages/1_Live.py | 2 +- pages/4_Dagoverzicht.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/pages/1_Live.py b/pages/1_Live.py index 9052a84..6687d04 100644 --- a/pages/1_Live.py +++ b/pages/1_Live.py @@ -45,7 +45,7 @@ def _basal_and_target_series(profile: dict, treatments: pd.DataFrame, window_sta dates = pd.date_range(window_start.normalize(), window_end.normalize(), freq="D") basal_parts, target_parts = [], [] for day_start in dates: - day_end = day_start + pd.Timedelta(days=1) + day_end = day_start + pd.Timedelta(1, unit="D") basal_parts.append(effective_basal_series(profile, treatments, day_start, day_end)) target_parts.append(target_glucose_series(profile, day_start, day_end)) diff --git a/pages/4_Dagoverzicht.py b/pages/4_Dagoverzicht.py index 6584a08..0bc6e5a 100644 --- a/pages/4_Dagoverzicht.py +++ b/pages/4_Dagoverzicht.py @@ -90,7 +90,7 @@ fig = make_subplots( for i, day in enumerate(dates, start=1): day_df = glucose[glucose["date"] == day] day_start = pd.Timestamp(day) - day_end = day_start + pd.Timedelta(days=1) + day_end = day_start + pd.Timedelta(1, unit="D") fig.add_hrect( y0=TARGET_LOW, y1=TARGET_HIGH, fillcolor=colors["band"], line_width=0, layer="below", row=i, col=1 -- 2.54.0