16

Data Handling - Part 4

• Several types of calculators are prohibited in exam halls. Refer to the file on Canvas for a list of permitted and prohibited calculators.

Measurements of Accuracy

• Calculate accuracy by determining the error rate of a set of replicate measurements.
• Error Rate is calculated as: | Observed Value - Actual Value | / Actual Value × 100.
• Example: for measurements of 174.9, 140.8, 190.8, 204.9, 172.2, 210, 156.4, 162.6 ng, with a true value of 174 ng, the error rate is 1.479%.

Measurements of Accuracy (Recovery)

• Recovery is determined by spiking sample matrices (serum, EDTA plasma, and heparin plasma) with a known amount of control analyte.
• The average percent recovery should fall between 80-120% for acceptable results.
• Example recovery ranges:
  • Serum (n=5): 92-101%, Average = 95%
  • EDTA plasma (n=5): 89-104%, Average = 98%
  • Heparin plasma (n=5): 86-104%, Average = 95%

Recovery (Immunoassay)

• In an immunoassay recovery experiment, calculate the percentage recovery.
• Two sample tubes are prepared, one spiked with analyte (Tube 1) and the other with water (Tube 2).
• Calculate the recovery using the formula:
  • % recovery = [(conc. tube 1) - (conc. tube 2)] x 100 / (conc. analyte added to tube 1)
• Example: Conc. analyte measured in tube 1 = 110 mg/ml, Conc. analyte measured in tube 2 = 0 mg/ml, Conc. analyte added =100 mg/ml.
  • % recovery = [(110 - 0) x 100] / 100 = 110%

Measurements of Accuracy (Linearity)

• Linearity is determined by spiking various sample matrices, making four two-fold serial dilutions, and measuring the average percent recovery.
• Example linear ranges in different sample matrices:
  • Serum (n=5): 1:2 (91-98%), 1:4 (82-96%), 1:8 (78-104%), 1:16 (83-92%)
  • EDTA plasma (n=5): 1:2 (93-105%), 1:4 (89-101%), 1:8 (88-97%), 1:16 (80-93%)
  • Heparin plasma (n=5): 1:2 (90-103%), 1:4 (96-105%), 1:8 (97-105%), 1:16 (96-105%)

Calculating Linearity

• Linearity is assessed using a regression equation, determining the relationship between bias and the reference values of the samples.
• The regression formula used is: y = ax + b, where y = bias value, a = slope, x = reference value, b= y-intercept.
• The bias is the difference between the sample measurement and the reference measurement.

Linear Curve Fitting

• Analyze immunoassay data using linear regression curve fitting, plotting concentration against assay readout (e.g., OD for ELISA).
• Use the standard equation: y = ax +b, where x is concentration, y is assay readout, a is slope, and b is y-intercept.
• The aim is to find the best slope and y-intercept that minimize the absolute distance between each data point and the curve (residuals).

Linear Regression

• A graphical visualization of linear regression. Observed data points are shown with fitted values and residuals.

Non-linear Curve Fitting (4-Parameter Logistic)

• Immunoassay data often produces an S-shaped sigmoidal curve.
• 4-Parameter Logistic (4PL) curve fitting is suitable for non-linear data, where the full range of the standard curve is evaluated, including the non-linear part.
• The underlying concept of summing the square of the residuals is still used, but the curve fit is optimized to match the S-shaped curve's specifics.

This quiz covers topics related to measurements of accuracy and recovery in data handling. You will learn how to calculate error rates and evaluate percent recovery for various sample matrices. Delve into specific examples to enhance your understanding of these critical concepts.