Establish high-accuracy AI governance technology to produce a local path

ogen-themed database.
1.Cooperate with the National Health Insurance (NHI) Administration to a
pply natural language processing (NLP) technology to develop a set of me
thods for managing pathogen data, which can automatically standardize t
he data format, and combine the expertise of infectious disease physician
s to make the standardized data meet the needs of professional fields.
2.Using NLP technology, the data uploaded from the hospital to the infor
mation system of the NHI Administration, it will automatically collects the
strain name, antibiotic name, antibiotic susceptibility and resistance value.
The data collection process includes the following steps:

(1)Data classification: Classify the uploaded data into two types: complete
and incomplete;
(2)Word tokenization: Use Regular Expression to separate the string of co
mpleted data into individual words or symbols;
(3)Extract key data: From the comparison results of the tokenized data, ext
ract the strain name, antibiotic name, antibiotic susceptibility and resistan
ce values;
(4) Look for the law of data: look for the law of pathogen resistance data,
e.g., [antibiotic name][abbreviation of antibiotic][antibiotic susceptibili
ty][resistance value];
(5) Compiling data: according to the law of key data, compile data sequen
tially;
(6) Data integration: output the pattern data and the aggregated data in d
ifferent Excel files.
Results:
1.High accuracy: Develop AI models to rectify pathogen-themed database
s. The correct rate of data compilation can reach 93.88%.
2.Save manpower-time:
(1)Non-blood bacterial data: It only takes about 65 hours to collect 240,00
0 pieces of data using this technology. It would take about 1,000 working
days to sort out these materials manually.
(2)Blood bacteria data: It only takes about 25 hours to collect 60,000 piece
s of data using this technology. It would take about 250 working days to s
ort out these m