Analyzing patient experiences using natural language processing: development and validation of the artificial intelligence patient reported experience measure (AI-PREM)

Table 1 Overview of the number of AI-PREM responses per sentiment

Questions	Number of patients N (%)	Average PEM scores of matched questions, ranging from 1 to 10 µ ± sd
Q1–Positive Negative	359 (67.2%) 26 (4.9%)	9.7 ± 0.9 8.1 ± 2.4**
Q2–Positive Negative	360 (67.4%) 31 (5.8%)	9.7 ± 0.7 7.7 ± 2.6**
Q3–Positive Negative	325 (60.9%) 40 (7.5%)	9.6 ± 1.1 8.3 ± 1.8*
Q4–Positive Negative	343 (64.2%) 39 (7.3%)	6.9 ± 1.7 6.4 ± 2.0
Q5–Positive Negative	121 (22.7%) 35 (6.6%)

The neutral responses are left out. Per category (question and sentiment), the average scores to the PEM questions that matched the AI-PREM questions are shown. P-value for the t-test for independent samples: * = p < 0.001, ** = p < 0.0001. AI-PREM: artificial intelligence patient reported experience measure. PEM: patient experience monitor. Q: question. sd: standard deviation

ISSN: 1472-6947