**2.3. Data analysis and result**

intensity via questionnaire including happiness, sadness, disgust, surprise, fear, and anger, from 0 (lowest intensity) to 8 (highest intensity). DSM-5 IGD criteria and online game film

Internet addiction (IA) is usually evaluated using self-reported questionnaires, scales, or interviews. Chen Internet Addiction Scale (CIAS) [4], a well-developed and widely applied scale, was adopted. CIAS consists of 26 items and is a self-assessment scale including five dimensions of Internet-related symptoms and problems, namely symptoms of compulsive use, of withdrawal, and of tolerance, and problems in interpersonal relationship, and in health/time management. People can be classified into high-risk IA (HIA) or low-risk IA (LIA). The nine IGD criteria were used in IGD experiment. The cut-point was five, and people with the score over five were IGD. Physiological signals including respiratory signals, ECG signals, and facial images were collected. Respiratory signals of thoracic movement (TM), abdominal movement (AM), and thoracoabdominal movement (TAM) were acquired using two respiratory belts with sampling rate at 1000 Hz (SS5LB, BIOPAC Systems, Inc., Goleta, USA). The belts were encircled at the level of armpit (TM, channel 1, C1) and navel (AM, channel 3, C3). Electrocardiography (ECG, with three disposable pregelled Ag/AgCl spot electrodes) was used with electrodes applied to the surface of a participant's skin. ECG signals were sampled at 1000 Hz and acquired using the DAQcard (USB 6218, National Instruments Corp., Austin, USA). Facial images were captured using a webcam (Logitech V-UBK45, USB 2.0, 10 fps, 640 × 480 resolution, Switzerland).

Sixty-eight participants (12 females and 56 males) aged between 20 and of 29 years were recruited from one university in Taiwan for IA experiment. None of the participants had bipolar or related disorders, depressive disorders, anxiety disorders, or agoraphobia. Participants were randomly divided into two groups, group 1: the emotional picture trial group, 34 participants (6 females and 28 males) aged between 19 and 25 years, and group 2: the emotional film trial group, 34 participants (6 females and 28 males) aged between 19 and 27 years. All participants filled out the CIAS, and were divided into the HIA group (n = 15 of group 1, n = 19 of group 2) or the LIA group (n = 19 of group 1, n = 15 of group 2). This study was approved by the Institutional Review Board of the National Taiwan University Hospital Hsinchu Branch

Fifty participants (14 females and 36 males) aged between 20 and 36 years were recruited from two universities in Taiwan for IGD experiment. The participants were divided into IGD & HIA group (n = 19, 4 females and 15 males, aged between 20 and 36) and non-IGD & LIA group (n = 21, 9 females and 12 males, aged between 20 and 28) using both CIAS and DSM-5 criteria. This study was approved by Research Ethics Committee for Human Subject Protection of National Chiao Tung University (Hsinchu, Taiwan) under the Project Number NCTU-REC-102-009.

The experimental procedure consisted of three phases. First, in phase 0, participants were seated in a comfortable chair and introduced to the experimental purpose and procedure, and then signed an informed consent form. Second, in phase 1, the physiological baseline including respiratory signals (TAM), ECG signals, and facial images of participants was measured

(Hsinchu, Taiwan) under the research Project Number 100IRB-32.

clips as emotional induction materials were adopted for IGD experiment.

**2.2. Experimental setup and procedure**

108 Drug Addiction

Data analysis method included statistics, such as correlation coefficient, T-test, and factorial ANOVA, and signal processing method for ECG and respiratory signals. ECG signals were processed and analyzed using an R-peak detection method. The RR intervals were transformed into an auto power spectrum, and the obtained HRV contained three frequency bands—high frequency (HF, 0.15–0.4 Hz), low frequency (LF, 0.04–0.15 Hz), and very low frequency (VLF, <0.04 Hz, excluding 0.00 HZ). Usually, HF indicates a parasympathetic nervous response and LF indicates a sympathetic nervous response. The RSA value was calculated as HF/((HF + LF)). Complementary Ensemble Empirical Mode Decomposition (CEEMD) which has been validated to be suitable for respiratory signals was applied to decompose respiratory signals [55]. CEEMD can acquire high frequency (HF, muscle contraction frequency), dominant frequency (DF, main respiratory frequency), and power of each respiratory signal. We are curious about the responses or reaction intensity (normalized power difference value) to several emotional stimuli from different time scale materials of IA abusers. The value of normalized power difference upon before and after the emotional inductions indicated the respiratory amplitude (emotional intensity), and the trend of difference implied the positive and negative respiratory responses. The differences in TAM, ECG (RSA) between HIA and LIA were examined.
