Page 77 - Taiwan Food and Drug Administration 2016 Annual Report

2016 ANNUAL
REPORT

2. Using Historical Data to Establish Future Prediction Models
(1) Ensure Data Comprehensiveness
In 2015, historical data gathered by the Five-Must information system of TFDA, namely the
automated management system for border inspections (Must-Declare), Product Distribution
Management System (Must-Test), Food Business Registration System (fadenbook, Must-
Register), inspection system (Must-Inspect), and traceability system (ftracebook, Must-
Trace), were integrated to undergo error proo?ng and data veri?cation. Once data quality and
accuracy had been validated, analysis of historical data from different systems was carried out
to achieve an overall understanding of food safety issues in Taiwan. Part II - Key Administrative Results Risk Assessment and Management

(2) Establishing Risk Prediction Indicators
Statistical analysis tools (such as R, SPSS, Modeler, and Tableau) were used to carry out data
analysis and data mining of historical data and information gathered by border inspections and
post-marketing sampling and tests of various products. Results were then used to establish
preliminary risk prediction indicators for food safety management. In 2015, a total of ?ve major
risk prediction indicators were established with analysis techniques to help screen and select
high-risk targets, namely: (1) number of inspected batches of imports; (2) net import weight; (3)
border sampling rate; (4) multi-year failure rate of testing; and (5) historical trends of test failure
rate. Regular discussion meetings were held with the relevant businesses to provide empirical
basis in stipulating source management policies.
(3) Establish an Early Warning Analysis Module for Risks
TFDA started using the established risk prediction indicators to build long-term systematic
risk warning and analysis modules to automatically select high-risk targets and achieve
the objectives of real-time monitoring. TFDA will also update information on a regular basis
and continue to evaluate and analyze the suitability of the module. Actual management
requirements will also be referenced to revise parameters of the analysis module to effectively
achieve the desired monitoring and early warning functions.

Section 3. Crisis Management System
Current Status
Courses were held in 2015 in order to help TFDA staff members improve their familiarity with
risk management and crisis handling concepts in order to respond to and handle emergencies.
Public opinions and information were reviewed on a daily basis. Regular post-market surveillance
was also conducted to maintain the safety of goods sold on the market for the general public.
Corresponding response systems to quickly handle and contain the issue are also in place in case
of actual emergencies. Once the incident is over, TFDA will formulate improvements to improve
crisis handling capabilities.
Policies and Outcomes

1. Organizing, Implementing, and Managing Risk Management Systems
(1) Risk Management and Crisis Handling Training
TFDA continued to provide training in risk management and crisis handling for fellow staff
members to strengthen their familiarity for the overall concepts of risk identi?cation and crisis
prevention. In 2015, a total of three training sessions for 190 attendants were held.

75