Table 9 Summary of LIME XAI description.

From: A comparative study on TB incidence and HIVTB coinfection using machine learning models on WHO global TB dataset

Dataset Features

Direction of contribution

Positive/negative contribution

Feature risk explanation

Size of prediction Probability

Feature Values

e_mort_100k

<= -0.46

very high positive

Low

0.43

-0.48

c_newinc_100k

<= -0.62

positive

Low

0.32

-0.68

cfr

<= -0.62

very high negative

high

0.21

-0.92

c_cdr

0.35 < c_cdr < = 0.72

slight positive

Low

0.07

0.72

e_pop_num

-0.21 < e_pop_num <=

negative

high

0.06

-0.18

e_tbhiv_prct

-0.67 < e_tbhiv_prct <=

very low positive

Low

0.04

-0.64