Real time biostatistics applied in acute myeloid leukemia / Biostatistica

1
Real time biostatistics applied in acute myeloid
leukemia / Biostatistica în timp real aplicata în
leucemia acuta mieloida

• This paper is partially supported by the Sectoral
  Operational Programme Human Resources
  Development, financed from the European Social
  Fund and by the Romanian Government under the
  contract number POSDRU/89/1.5/S/60782.

This work was supported by CNCS UEFISCDI,
project number PN-II-ID-WE-2012-4-162/2012
2
Introduction

• Acute myeloid leukemia
• Malignant disease of the myeloid line
• Accumulation of abnormal white blood cells in the
  bone marrow which interferes with the production
  of normal blood cells
• Symptoms fatigue, shortness of breath, easy
  bruising and bleeding, increased risk of infection

3
Introduction

• Acute myeloid leukemia
• The disease can be fatal within weeks or months
  without treatment
• Standardized treatment and diagnosis techniques
• Complex patient management
• Rare disease ( 1,2 of cancer deaths in US)
• Incidence is increasing as the population ages

4
Introduction

• LAM clinical studies
• There is a high variability in the recorded
  parameters during the evolution of the disease.
• Because of the small number of patient any
  mistake in data collection can reduce the level
  of significance of the study results
• The results of the study can reveal a treatment
  misconduct / how the treatment is received by the
  patients

5
Goal

• Our aim was to create a software able to
  automatize as much as possible the process of
  data collection, data quality management and
  statistical processing
• This should reduce the chance of error in data
  collection and would offer a realtime
  biostatistical overview of the results
• The results of statistical processing will be
  available during the study, not only at the end,
  helping the physician to manage his patients

6
Material and Methods

• Programming design
• The software is written in C, and is running on
  a Linux system
• Database backend uses SQLite3 library
• The graphical interface has been written using
  wxWidgets
• All components are crossplatform able, so the
  source code can be adapted to compile and run on
  Windows and Mac OSX platforms as well

7
Material and Methods

• Programming design
• Statistical processing has been accomplished by
  using R software written at the University of
  Auckland
• R is accepted by the academic community as a
  standard statistical processing system of the
  same level as SAS and SPSS
• R is used automatically by the software to
  analyze the data without any user intervention

8
Material and Methods

• Advantages of this approach
• Portable system
• Very low overhead
• User friendly interface
• Open source system
• Early error handling
• Early data processing

9
Material and Methods

• The patients with AML selected and included in
  this study were hospitalized in the Hematological
  Department, County Emergency Clinical Hospital
  Tirgu Mures.

10
Results
Fig. 1. Real time analysis interface with
automatic survival curve (Kaplan Mayer) with 95
Confidence Interval (right window) and log rank
test (left window)
11
Results
Fig. 2. Data entry interface
12
Results
Fig. 3. Menu "General Settings" for establishing
threshold values for laboratory measurements
13
Results
Fig. 4. Selecting the interest variable.
14
Results
Fig. 5. Real time p value for log rank test
15
Results
Fig. 6. Real time descriptive statistics.
16
Results
Fig. 7. Real time Survival curves (log rank test)
17
Discussions

• At this time, most of the medical studies do not
  use such an automated system for data collection,
  validation and statistical processing or they are
  using simple forms for collecting data in
  Microsoft Access databases or Excel spreadsheets
  without any processing or complex validation.
• The statistical processing is done only at the
  end of the study. The potential problems of a
  specific treatment are detected too late for the
  patients involved in the study.

18
Discussions

• Using such a software is very useful in medical
  studies, reducing the potential errors
• The software developed and presented above,
  automatize the data quality control process and
  provides statistical analysis on the fly
• The survival curve is a mirror of the treatment
  success. The changes of the survival curve during
  the data collection can help the physician in
  detecting treatment errors.

19
Conclusions

• Data quality management issues are automatically
  solved using the software we proposed
• Statistical parameters are calculated in real
  time, providing snapshots of the statistical
  results throughout data entry process. The end of
  data collection coincides with a final report of
  the study.

20
Thank you for your attention!