Skip to contents

Line Charts for Clinical Time Series and Trend Analysis

ClinicoPath Package

2025-07-13

Source: vignettes/jjstatsplot-30-linechart-comprehensive.Rmd
jjstatsplot-30-linechart-comprehensive.Rmd

Introduction to Line Charts in Clinical Research

What are Line Charts?

Line charts are fundamental visualization tools that display data points connected by straight line segments, making them ideal for showing trends and changes over time or ordered categories. In clinical and pathological research, line charts are essential for:

  • Longitudinal studies: Tracking patient outcomes over time
  • Treatment monitoring: Visualizing response to interventions
  • Biomarker evolution: Following laboratory values or disease markers
  • Quality metrics: Monitoring healthcare performance indicators
  • Dose-response relationships: Understanding therapeutic relationships

Key Advantages

  1. Trend Identification: Easily spot increasing, decreasing, or stable patterns
  2. Comparative Analysis: Compare multiple groups or treatments simultaneously
  3. Time Series Visualization: Perfect for longitudinal and follow-up studies
  4. Statistical Integration: Combine with trend lines, confidence intervals, and correlations
  5. Clinical Communication: Intuitive format for presenting to clinicians and patients

When to Use Line Charts

Ideal Clinical Scenarios

  • Laboratory Values Over Time: Hemoglobin, tumor markers, inflammatory indices
  • Treatment Response Monitoring: Blood pressure, pain scores, functional assessments
  • Disease Progression: Tumor size, cognitive decline, symptom severity
  • Quality Improvement: Infection rates, readmission rates, patient satisfaction
  • Dose-Response Studies: Efficacy or toxicity versus drug concentration

Data Requirements

  • X-axis: Time points, visit numbers, or ordered categories
  • Y-axis: Continuous numeric outcomes (laboratory values, scores, measurements)
  • Optional Grouping: Treatment arms, patient populations, hospital units
  • Minimum Data: At least 3 time points for meaningful trend analysis

Statistical Background

Trend Analysis Methods

Line charts can incorporate linear regression to identify overall trends:

Y=β0+β1×Time+ϵY = \beta_0 + \beta_1 \times Time + \epsilon

Where: - β1\beta_1 represents the rate of change per time unit - Positive β1\beta_1: improving/increasing trend - Negative β1\beta_1: declining/decreasing trend

Correlation Analysis

The relationship between time and outcome can be quantified using:

  • Pearson correlation: For linear relationships
  • Spearman correlation: For monotonic (rank-based) relationships
  • R-squared: Proportion of variance explained by time

Confidence Intervals

Line charts can display uncertainty around trends: - Standard error bands: Show precision of the line - Confidence intervals: Typically 95% CI around the trend - Prediction intervals: Bounds for future observations

Clinical Interpretation Guidelines

Trend Magnitude

Clinical Interpretation of Trend Magnitude
Change.per.Unit.Time Clinical.Significance Clinical.Action Examples
< 5% Minimal Monitor Lab variation
5-15% Small Consider intervention Modest improvement
15-30% Moderate Likely clinically meaningful Notable response
> 30% Large Definite clinical impact Major clinical change

Correlation Strength

Correlation Strength Interpretation
Correlation..r. Strength Clinical.Utility R…Variance.Explained.
0.0 - 0.1 Negligible No trend < 1%
0.1 - 0.3 Weak Minimal trend 1-9%
0.3 - 0.5 Moderate Noticeable pattern 9-25%
0.5 - 0.7 Strong Clear trend 25-49%
0.7 - 1.0 Very Strong Highly predictable 49-100%

Clinical Examples

Example 1: Hemoglobin Monitoring During Cancer Treatment

This example demonstrates monitoring anemia treatment in cancer patients receiving erythropoietin (EPO) therapy.

# Create realistic hemoglobin data
set.seed(123)
n_patients <- 90
n_visits <- 6
visit_weeks <- c(0, 2, 4, 8, 12, 16)

# Generate patient data
hemoglobin_data <- expand.grid(
  patient_id = paste0("P", sprintf("%03d", 1:n_patients)),
  visit_week = visit_weeks
) %>%
  left_join(
    data.frame(
      patient_id = paste0("P", sprintf("%03d", 1:n_patients)),
      treatment_group = sample(c("Control", "EPO_Low", "EPO_High"), n_patients, 
                              replace = TRUE, prob = c(0.3, 0.4, 0.3)),
      baseline_hgb = rnorm(n_patients, mean = 9.5, sd = 1.2)
    ),
    by = "patient_id"
  ) %>%
  arrange(patient_id, visit_week)

# Generate hemoglobin trajectories
hemoglobin_data$hemoglobin_g_dl <- NA

for (i in 1:nrow(hemoglobin_data)) {
  week <- hemoglobin_data$visit_week[i]
  treatment <- hemoglobin_data$treatment_group[i]
  baseline <- hemoglobin_data$baseline_hgb[i]
  
  # Treatment-specific trends
  if (treatment == "Control") {
    trend <- -0.03 * week + rnorm(1, 0, 0.3)
  } else if (treatment == "EPO_Low") {
    trend <- 0.08 * week + rnorm(1, 0, 0.25)
  } else {
    trend <- 0.12 * week + rnorm(1, 0, 0.2)
  }
  
  hemoglobin_data$hemoglobin_g_dl[i] <- baseline + trend
}

# Ensure realistic ranges
hemoglobin_data$hemoglobin_g_dl <- pmax(6, pmin(16, hemoglobin_data$hemoglobin_g_dl))
hemoglobin_data$hemoglobin_g_dl <- round(hemoglobin_data$hemoglobin_g_dl, 1)

# Display data structure
str(hemoglobin_data)
#> 'data.frame':    540 obs. of  5 variables:
#>  $ patient_id     : chr  "P001" "P001" "P001" "P001" ...
#>  $ visit_week     : num  0 2 4 8 12 16 0 2 4 8 ...
#>  $ treatment_group: chr  "EPO_Low" "EPO_Low" "EPO_Low" "EPO_Low" ...
#>  $ baseline_hgb   : num  8.15 8.15 8.15 8.15 8.15 ...
#>  $ hemoglobin_g_dl: num  8.4 7.9 8.7 9.3 8.8 9.6 8.9 8.5 8.4 8.3 ...
#>  - attr(*, "out.attrs")=List of 2
#>   ..$ dim     : Named int [1:2] 90 6
#>   .. ..- attr(*, "names")= chr [1:2] "patient_id" "visit_week"
#>   ..$ dimnames:List of 2
#>   .. ..$ patient_id: chr [1:90] "patient_id=P001" "patient_id=P002" "patient_id=P003" "patient_id=P004" ...
#>   .. ..$ visit_week: chr [1:6] "visit_week= 0" "visit_week= 2" "visit_week= 4" "visit_week= 8" ...
head(hemoglobin_data, 12)
#>    patient_id visit_week treatment_group baseline_hgb hemoglobin_g_dl
#> 1        P001          0         EPO_Low     8.152270             8.4
#> 2        P001          2         EPO_Low     8.152270             7.9
#> 3        P001          4         EPO_Low     8.152270             8.7
#> 4        P001          8         EPO_Low     8.152270             9.3
#> 5        P001         12         EPO_Low     8.152270             8.8
#> 6        P001         16         EPO_Low     8.152270             9.6
#> 7        P002          0         Control     9.016538             8.9
#> 8        P002          2         Control     9.016538             8.5
#> 9        P002          4         Control     9.016538             8.4
#> 10       P002          8         Control     9.016538             8.3
#> 11       P002         12         Control     9.016538             8.5
#> 12       P002         16         Control     9.016538             8.1

Running Line Chart Analysis 

# In jamovi, you would select:
# Analyses → JJStatsPlot → Line Chart

# Programmatically (if running in R):
hemoglobin_result <- linechart(
  data = hemoglobin_data,
  xvar = "visit_week",
  yvar = "hemoglobin_g_dl",
  groupby = "treatment_group",
  confidence = TRUE,
  trendline = TRUE,
  points = TRUE,
  refline = 12,
  reflineLabel = "Normal Lower Limit",
  xlabel = "Weeks Since Treatment Start",
  ylabel = "Hemoglobin (g/dL)",
  title = "Hemoglobin Response to EPO Treatment",
  colorPalette = "clinical"
)

Expected Clinical Results

  1. Visual Patterns:
    • Control group: Slight decline or stable low values
    • EPO Low dose: Gradual improvement over time
    • EPO High dose: Steeper improvement, reaching normal range
  2. Statistical Analysis:
    • Strong positive correlation for EPO groups (r > 0.7)
    • Weak negative correlation for control group
    • Significant trend p-values for treatment groups
  3. Clinical Interpretation:
    • EPO therapy effectively treats cancer-related anemia
    • Higher doses provide faster hemoglobin recovery
    • Reference line helps identify patients reaching normal range

Example 2: Blood Pressure Response to Interventions

This example shows hypertension management comparing lifestyle interventions, medications, and combined therapy.

# Create blood pressure monitoring data
set.seed(456)
n_subjects <- 75
n_months <- 8
months <- 0:(n_months-1)

bp_data <- expand.grid(
  subject_id = paste0("BP", sprintf("%03d", 1:n_subjects)),
  month = months
) %>%
  left_join(
    data.frame(
      subject_id = paste0("BP", sprintf("%03d", 1:n_subjects)),
      intervention = sample(c("Lifestyle", "Medication", "Combined"), n_subjects, 
                           replace = TRUE, prob = c(0.35, 0.35, 0.3)),
      baseline_sbp = rnorm(n_subjects, mean = 155, sd = 15)
    ),
    by = "subject_id"
  ) %>%
  arrange(subject_id, month)

# Generate systolic BP trajectories
bp_data$systolic_bp <- NA

for (i in 1:nrow(bp_data)) {
  month <- bp_data$month[i]
  intervention <- bp_data$intervention[i]
  baseline <- bp_data$baseline_sbp[i]
  
  # Intervention-specific effects
  if (intervention == "Lifestyle") {
    reduction <- 2 * sqrt(month) + rnorm(1, 0, 2)
  } else if (intervention == "Medication") {
    reduction <- ifelse(month <= 3, 8 * (month / 3), 8 + 2 * ((month - 3) / 5)) + rnorm(1, 0, 3)
  } else {
    reduction <- ifelse(month <= 2, 6 * (month / 2), 6 + 8 * ((month - 2) / 6)) + rnorm(1, 0, 2)
  }
  
  bp_data$systolic_bp[i] <- baseline - reduction
}

# Ensure realistic ranges
bp_data$systolic_bp <- pmax(90, pmin(200, bp_data$systolic_bp))
bp_data$systolic_bp <- round(bp_data$systolic_bp)

# Display summary
cat("Blood Pressure Study Summary:\n")
#> Blood Pressure Study Summary:
cat("Subjects:", length(unique(bp_data$subject_id)), "\n")
#> Subjects: 75
cat("Interventions:", paste(unique(bp_data$intervention), collapse = ", "), "\n")
#> Interventions: Medication, Combined, Lifestyle
cat("Baseline SBP range:", min(bp_data[bp_data$month == 0, "systolic_bp"]), "-", 
    max(bp_data[bp_data$month == 0, "systolic_bp"]), "mmHg\n")
#> Baseline SBP range: 113 - 192 mmHg
cat("Final SBP range:", min(bp_data[bp_data$month == max(bp_data$month), "systolic_bp"]), "-", 
    max(bp_data[bp_data$month == max(bp_data$month), "systolic_bp"]), "mmHg\n")
#> Final SBP range: 101 - 187 mmHg

Blood Pressure Analysis Setup 

bp_result <- linechart(
  data = bp_data,
  xvar = "month",
  yvar = "systolic_bp",
  groupby = "intervention",
  confidence = TRUE,
  trendline = TRUE,
  refline = 140,
  reflineLabel = "Hypertension Threshold",
  xlabel = "Months Since Intervention Start",
  ylabel = "Systolic Blood Pressure (mmHg)",
  title = "Blood Pressure Response to Different Interventions",
  colorPalette = "colorblind",
  theme = "publication"
)

Expected Clinical Findings

  1. Intervention Comparisons:
    • Lifestyle: Gradual, sustained reduction
    • Medication: Rapid initial improvement, then plateau
    • Combined: Best overall reduction combining both benefits
  2. Clinical Milestones:
    • Time to reach target BP (<140 mmHg)
    • Magnitude of reduction from baseline
    • Sustainability of improvements

Example 3: Biomarker Response to Immunotherapy

This example tracks inflammatory biomarkers (IL-6, TNF-α) during cancer immunotherapy to assess treatment response.

# Create biomarker immunotherapy data
set.seed(789)
n_patients <- 45
visit_days <- c(0, 21, 42, 84, 126, 168)

biomarker_data <- expand.grid(
  patient_id = paste0("IMM", sprintf("%03d", 1:n_patients)),
  visit_day = visit_days
) %>%
  left_join(
    data.frame(
      patient_id = paste0("IMM", sprintf("%03d", 1:n_patients)),
      response_type = sample(c("Responder", "Non_Responder", "Progressive"), n_patients,
                            replace = TRUE, prob = c(0.4, 0.35, 0.25)),
      baseline_il6 = rlnorm(n_patients, meanlog = 2, sdlog = 0.8)
    ),
    by = "patient_id"
  ) %>%
  arrange(patient_id, visit_day)

# Generate IL-6 trajectories
biomarker_data$il6_pg_ml <- NA

for (i in 1:nrow(biomarker_data)) {
  day <- biomarker_data$visit_day[i]
  response <- biomarker_data$response_type[i]
  baseline <- biomarker_data$baseline_il6[i]
  
  # Response-specific patterns
  if (response == "Responder") {
    factor <- exp(-0.003 * day) * (1 + rnorm(1, 0, 0.2))
  } else if (response == "Non_Responder") {
    factor <- 1 + 0.1 * sin(day / 50) * (1 + rnorm(1, 0, 0.3))
  } else {
    factor <- 1 + 0.002 * day * (1 + rnorm(1, 0, 0.4))
  }
  
  biomarker_data$il6_pg_ml[i] <- baseline * factor
}

# Ensure positive values
biomarker_data$il6_pg_ml <- pmax(0.1, pmin(100, biomarker_data$il6_pg_ml))
biomarker_data$il6_pg_ml <- round(biomarker_data$il6_pg_ml, 2)

# Calculate summary by response type
response_summary <- biomarker_data %>%
  filter(visit_day %in% c(0, 168)) %>%
  select(patient_id, response_type, visit_day, il6_pg_ml) %>%
  tidyr::pivot_wider(names_from = visit_day, values_from = il6_pg_ml, names_prefix = "day_") %>%
  mutate(percent_change = round(100 * (day_168 - day_0) / day_0, 1)) %>%
  group_by(response_type) %>%
  summarise(
    n = n(),
    mean_baseline = round(mean(day_0, na.rm = TRUE), 1),
    mean_final = round(mean(day_168, na.rm = TRUE), 1),
    mean_change = round(mean(percent_change, na.rm = TRUE), 1),
    .groups = 'drop'
  )

print(response_summary)
#> # A tibble: 3 × 5
#>   response_type     n mean_baseline mean_final mean_change
#>   <chr>         <int>         <dbl>      <dbl>       <dbl>
#> 1 Non_Responder    18           9.4        9.2        -2.3
#> 2 Progressive       5           8         10.8        37.2
#> 3 Responder        22           9.6        7.4       -29

Biomarker Analysis 

biomarker_result <- linechart(
  data = biomarker_data,
  xvar = "visit_day",
  yvar = "il6_pg_ml", 
  groupby = "response_type",
  confidence = TRUE,
  smooth = TRUE,  # Use smoothed lines for biomarker data
  xlabel = "Days Since Treatment Start",
  ylabel = "IL-6 (pg/mL)",
  title = "IL-6 Response Patterns During Immunotherapy",
  colorPalette = "viridis"
)

Clinical Biomarker Interpretation

  1. Response Patterns:
    • Responders: Progressive decline in inflammatory markers
    • Non-responders: Stable levels with minimal change
    • Progressive disease: Increasing inflammatory activity
  2. Predictive Value:
    • Early biomarker changes may predict long-term response
    • Useful for treatment modification decisions
    • Complement to imaging and clinical assessments

Advanced Line Chart Features

Confidence Intervals and Uncertainty

Standard Error vs. Confidence Intervals 

# Line chart with confidence intervals
result <- linechart(
  data = clinical_data,
  xvar = "time_point",
  yvar = "biomarker_level",
  groupby = "treatment",
  confidence = TRUE,     # Shows 95% CI around lines
  smooth = FALSE,        # Linear vs. smoothed confidence bands
  theme = "publication"
)

Clinical Applications: - Uncertainty quantification: Show precision of measurements - Group comparisons: Overlapping CI suggests similar trends - Sample size effects: Wider CI with smaller groups - Statistical significance: Non-overlapping CI indicates differences

Reference Lines and Clinical Thresholds

Normal Ranges and Target Values 

# Add clinical reference lines
result <- linechart(
  data = lab_data,
  xvar = "visit_week", 
  yvar = "hemoglobin_g_dl",
  refline = 12,                    # Normal hemoglobin threshold
  reflineLabel = "Normal Range",   # Clear clinical labeling
  title = "Hemoglobin Monitoring with Clinical Targets"
)

Common Clinical References: - Laboratory normal ranges: Hemoglobin, liver enzymes, kidney function - Blood pressure targets: <140/90 mmHg for hypertension - Pain score thresholds: ≤3/10 for adequate pain control - Quality metrics: Infection rates, readmission benchmarks

Statistical Trend Analysis

Linear vs. Non-linear Trends 

# Compare linear and smoothed trends
linear_result <- linechart(
  data = longitudinal_data,
  xvar = "time",
  yvar = "outcome", 
  trendline = TRUE,    # Linear regression line
  smooth = FALSE       # Straight line connections
)

smooth_result <- linechart(
  data = longitudinal_data,
  xvar = "time",
  yvar = "outcome",
  smooth = TRUE,       # LOESS smoothing
  confidence = TRUE    # Smoothed confidence bands
)

When to Use Each: - Linear trends: Constant rate of change, dose-response relationships - Smoothed trends: Complex patterns, seasonal effects, non-linear responses

Clinical Applications by Specialty

Oncology Applications

Treatment Response Monitoring

Oncology Line Chart Applications
Clinical.Scenario X.axis.Variable Y.axis.Variable Grouping.Variable
Tumor marker trends Months since diagnosis PSA, CA-125, CEA levels Tumor stage/grade
Chemotherapy response Treatment cycles Tumor size (RECIST) Treatment regimen
Quality of life during treatment Weeks since treatment start EORTC QLQ scores Performance status
Biomarker-guided therapy Days since targeted therapy PD-L1, circulating tumor DNA Biomarker status
Survivorship monitoring Years post-treatment Late effects scores Treatment modality

Key Clinical Insights

  • Early response prediction: Biomarker changes predict long-term outcomes
  • Treatment optimization: Adjust therapy based on trends
  • Quality of life: Balance efficacy with patient-reported outcomes

Cardiology Applications

Cardiovascular Risk Management

Cardiology Monitoring Applications
Clinical.Scenario Outcome.Measures Clinical.Targets Typical.Timeline
Blood pressure management SBP/DBP (mmHg) <140/90 mmHg Months to years
Lipid control monitoring LDL cholesterol (mg/dL) <70 mg/dL (high risk) 3-6 month intervals
Heart failure progression EF%, BNP levels Stable/improved EF Quarterly assessments
Post-MI recovery Exercise capacity, LVEF Return to baseline function 6-12 months post-MI
Device monitoring ICD/pacemaker parameters Appropriate device function Continuous monitoring

Critical Care Applications

ICU Monitoring and Outcomes

Critical Care Line Chart Applications
Monitoring.Domain Key.Variables Time.Resolution Clinical.Goals
Hemodynamic stability MAP, cardiac output Hours to days Hemodynamic optimization
Respiratory function P/F ratio, PEEP levels Hourly to daily Weaning from ventilator
Organ function recovery Creatinine, liver enzymes Daily assessments Organ recovery
Sedation management RASS scores, delirium Shift-based scoring Optimal sedation
Nutritional status Albumin, weight changes Daily to weekly Adequate nutrition

Best Practices for Clinical Line Charts

Data Preparation Guidelines

1. Time Variable Standardization 

# Standardize time points for multi-site studies
clinical_data <- clinical_data %>%
  mutate(
    # Convert visit dates to study days
    study_day = as.numeric(visit_date - enrollment_date),
    
    # Group irregular visits into windows
    visit_window = case_when(
      study_day <= 7 ~ "Week 1",
      study_day <= 28 ~ "Month 1", 
      study_day <= 84 ~ "Month 3",
      study_day <= 168 ~ "Month 6",
      TRUE ~ "Later"
    ),
    
    # Handle missing values appropriately
    outcome_clean = ifelse(is.na(lab_value), NA, lab_value)
  )

2. Missing Data Strategies 

# Handle missing data patterns
missing_analysis <- clinical_data %>%
  group_by(patient_id) %>%
  summarise(
    n_visits = n(),
    n_missing = sum(is.na(primary_outcome)),
    last_visit = max(study_day, na.rm = TRUE),
    dropout_pattern = case_when(
      n_missing == 0 ~ "Complete",
      n_missing < n_visits/2 ~ "Sporadic missing",
      last_visit < 168 ~ "Early dropout",
      TRUE ~ "Late dropout"
    )
  )

# Visualize missing patterns
missing_plot <- linechart(
  data = clinical_data,
  xvar = "study_day",
  yvar = "primary_outcome",
  groupby = "dropout_pattern",
  title = "Outcome by Missing Data Pattern"
)

3. Outlier Management 

# Identify and handle outliers
clinical_data <- clinical_data %>%
  group_by(visit_week) %>%
  mutate(
    # Calculate z-scores within time points
    z_score = abs(scale(lab_value)[,1]),
    
    # Flag extreme outliers
    outlier_flag = z_score > 3,
    
    # Winsorize extreme values
    lab_value_winsorized = case_when(
      z_score > 3 ~ quantile(lab_value, 0.95, na.rm = TRUE),
      z_score < -3 ~ quantile(lab_value, 0.05, na.rm = TRUE),
      TRUE ~ lab_value
    )
  ) %>%
  ungroup()

Visualization Design Principles

1. Color and Theme Selection 

# Clinical publication-ready themes
publication_chart <- linechart(
  data = clinical_data,
  xvar = "time_point",
  yvar = "biomarker",
  groupby = "treatment",
  colorPalette = "colorblind",  # Accessible to colorblind readers
  theme = "publication",        # Clean, professional appearance
  xlabel = "Time (weeks)",
  ylabel = "Biomarker Level (units)",
  title = "Treatment Response Over Time"
)

# Specialty-specific color schemes
cardiology_chart <- linechart(
  data = bp_data,
  colorPalette = "clinical",    # Red/blue for medical contexts
  refline = 140,
  reflineLabel = "Hypertension Threshold"
)

2. Axis and Label Optimization 

# Optimize for clinical communication
optimized_chart <- linechart(
  data = clinical_data,
  xvar = "visit_month",
  yvar = "hemoglobin_g_dl",
  
  # Clear, clinical labels
  xlabel = "Months Since Treatment Initiation",
  ylabel = "Hemoglobin (g/dL)",
  title = "Hemoglobin Response to Iron Supplementation",
  
  # Reference lines for clinical context
  refline = 12,
  reflineLabel = "Normal Lower Limit (Women)",
  
  # Appropriate size for presentations
  width = 800,
  height = 600
)

Statistical Reporting Standards

1. Correlation and Trend Reporting 

# Extract statistical results for reporting
trend_analysis <- linechart(
  data = longitudinal_data,
  xvar = "time",
  yvar = "outcome",
  trendline = TRUE
)

# Report format for publications:
# "Hemoglobin levels showed a significant positive correlation with time 
# (r = 0.78, 95% CI: 0.65-0.87, p < 0.001), with an average increase 
# of 0.45 g/dL per month (95% CI: 0.32-0.58)."

2. Group Comparison Reporting 

# Compare trends between groups
group_analysis <- linechart(
  data = treatment_data,
  xvar = "weeks",
  yvar = "pain_score", 
  groupby = "treatment_arm",
  trendline = TRUE
)

# Report format:
# "Pain scores decreased significantly in both treatment groups over 12 weeks. 
# The experimental group showed a steeper decline (slope = -0.8 points/week, 
# r = -0.85, p < 0.001) compared to control (slope = -0.3 points/week, 
# r = -0.62, p = 0.003)."

Troubleshooting Common Issues

1. Irregular Time Points

Problem: Clinical visits don’t occur at regular intervals

Solutions:

# Option 1: Use actual visit times
irregular_chart <- linechart(
  data = visit_data,
  xvar = "actual_visit_day",  # Use actual times
  yvar = "lab_result",
  smooth = TRUE              # Smooth over irregular intervals
)

# Option 2: Standardize to visit windows
standardized_data <- visit_data %>%
  mutate(
    visit_period = case_when(
      actual_visit_day <= 10 ~ "Baseline",
      actual_visit_day <= 35 ~ "Month 1",
      actual_visit_day <= 95 ~ "Month 3",
      TRUE ~ "Month 6"
    )
  )

window_chart <- linechart(
  data = standardized_data,
  xvar = "visit_period",
  yvar = "lab_result"
)

2. Highly Variable Data

Problem: Large fluctuations obscure underlying trends

Solutions:

# Use smoothed lines for noisy data
smooth_chart <- linechart(
  data = noisy_data,
  xvar = "time",
  yvar = "variable_outcome",
  smooth = TRUE,             # LOESS smoothing
  confidence = TRUE,         # Show uncertainty
  points = FALSE             # Hide individual points
)

# Alternative: Moving averages
smoothed_data <- noisy_data %>%
  arrange(patient_id, time_point) %>%
  group_by(patient_id) %>%
  mutate(
    moving_avg = zoo::rollmean(outcome, k = 3, fill = NA, align = "center")
  ) %>%
  ungroup()

3. Multiple Y-Variables

Problem: Want to show multiple related outcomes

Solutions:

# Option 1: Standardize to z-scores
standardized_data <- clinical_data %>%
  mutate(
    sbp_z = as.numeric(scale(systolic_bp)),
    dbp_z = as.numeric(scale(diastolic_bp))
  ) %>%
  tidyr::pivot_longer(
    cols = c(sbp_z, dbp_z),
    names_to = "bp_type",
    values_to = "z_score"
  )

multi_outcome_chart <- linechart(
  data = standardized_data,
  xvar = "visit_month",
  yvar = "z_score",
  groupby = "bp_type",
  ylabel = "Standardized Blood Pressure (Z-score)"
)

# Option 2: Separate charts with same scale
sbp_chart <- linechart(data = bp_data, xvar = "month", yvar = "systolic_bp")
dbp_chart <- linechart(data = bp_data, xvar = "month", yvar = "diastolic_bp")

4. Small Sample Sizes

Problem: Few patients or time points

Solutions:

# Emphasize individual trajectories
small_sample_chart <- linechart(
  data = pilot_data,
  xvar = "time",
  yvar = "outcome", 
  points = TRUE,           # Show all data points
  confidence = FALSE,      # Don't show CI with small n
  smooth = FALSE,          # Use straight lines
  title = "Individual Patient Trajectories (n=8)"
)

# Consider spaghetti plots for individual patients
individual_chart <- linechart(
  data = pilot_data,
  xvar = "time",
  yvar = "outcome",
  groupby = "patient_id",  # Each patient as separate line
  theme = "minimal"
)

Publication and Presentation Guidelines

Journal Submission Standards

Figure Preparation Checklist 

# Publication-ready line chart
publication_figure <- linechart(
  data = study_data,
  xvar = "time_weeks",
  yvar = "primary_endpoint",
  groupby = "treatment_arm",
  
  # Professional appearance
  theme = "publication",
  colorPalette = "colorblind",
  
  # Appropriate sizing
  width = 800,   # Suitable for 2-column format
  height = 600,
  
  # Clear labeling
  xlabel = "Time (weeks)",
  ylabel = "Primary Endpoint (units)",
  title = "",    # Usually caption provides title
  
  # Statistical elements
  confidence = TRUE,
  trendline = TRUE
)

Essential Figure Elements

  1. Clear axes labels with units
  2. Legible font sizes (minimum 10pt)
  3. Colorblind-friendly palettes
  4. Statistical information (p-values, confidence intervals)
  5. Sample sizes for each group and time point
  6. Reference lines for clinical thresholds

Caption Writing Guidelines 

Figure 1. Treatment response over time by intervention group. 
Line chart showing mean (±95% CI) primary endpoint values over 
24 weeks of follow-up. Control group (n=45, blue), Intervention 
group (n=43, red). Dotted horizontal line indicates clinical 
significance threshold (threshold value). P-values from linear 
mixed-effects models comparing slopes between groups.

Presentation Best Practices

1. Audience-Specific Design 

# For clinical audiences
clinical_presentation <- linechart(
  data = clinical_data,
  colorPalette = "clinical",
  theme = "classic",
  
  # Larger fonts for projection
  width = 1000,
  height = 700,
  
  # Clear reference lines
  refline = clinical_threshold,
  reflineLabel = "Clinical Target"
)

# For research conferences
research_presentation <- linechart(
  data = research_data,
  theme = "publication",
  confidence = TRUE,        # Show statistical rigor
  trendline = TRUE,
  colorPalette = "viridis"  # Professional color scheme
)

2. Interactive Elements

For presentations where interaction is possible:

  • Hover information: Display exact values and sample sizes
  • Group toggle: Allow showing/hiding specific treatment groups
  • Time range selection: Focus on specific periods of interest
  • Statistical overlay: Toggle trend lines and confidence intervals

Conclusion

Line charts are powerful tools for clinical research visualization that enable:

Key Clinical Applications

  • Longitudinal Analysis: Track patient outcomes over time
  • Treatment Monitoring: Assess intervention effectiveness
  • Biomarker Evolution: Follow disease progression markers
  • Quality Improvement: Monitor healthcare performance metrics
  • Comparative Research: Compare treatments or populations

Statistical Capabilities

  • Trend Analysis: Linear and non-linear pattern detection
  • Correlation Assessment: Quantify time-outcome relationships
  • Confidence Intervals: Display uncertainty and precision
  • Group Comparisons: Statistical comparison of trends

Best Practices Summary

  1. Design for audience: Clinical vs. research presentations
  2. Include clinical context: Reference lines and thresholds
  3. Handle missing data: Transparent reporting of dropout patterns
  4. Statistical rigor: Report correlations, confidence intervals, and p-values
  5. Professional appearance: Publication-ready formatting

Future Directions

  • Real-time monitoring: Integration with electronic health records
  • Predictive modeling: Combine trends with machine learning
  • Patient-specific trajectories: Personalized medicine applications
  • Multi-outcome visualization: Simultaneous monitoring of related endpoints

The linechart function in ClinicoPath provides researchers and clinicians with a comprehensive tool for creating publication-quality visualizations that effectively communicate temporal patterns and treatment effects in clinical data.

For more information about ClinicoPath visualization tools and clinical data analysis, visit the package documentation and tutorials.