Analysis of time series of particle size distributions in nano exposure assessment

Real-time exposure measurements to nano-sized particles may result in large amounts of time series data on particle size and total number concentration. Analysis of the particle size distribution have thus far been limited to either graphical analysis of the distribution over time or an evaluation of the mode over time. For large time series data, graphical analysis of distributions is complicated and an assessment of the mode ignores the important aspect of the variance in particle size. A statistical method of analysis is proposed that overcomes those problems, based on a multilevel modeling approach and assuming a lognormal model for the particle size distribution. Two empirical examples illustrate the advantages of the proposed model, showing that useful summaries and inferences can be obtained, even for large data sets. The model thus provides a tool for practitioners to deal with large amounts of particle size distribution data obtained from real-time nano measurement devices.