Violence against women is a problem faced in several ways, in various societies; however, the introduction of computational tools is something still little explored in this confrontation. Thus, it is necessary to invest in researches that bring technological development closer to the prevention, discovery, and combat of this form of violence. This paper presents the Women's Health Observer Tool (WHOT) that helps to build psychobehavioral profiles of women victims of violence, based on three features: i) recognition of facial expressions to infer emotions; ii) provision of digital questionnaires on intimate partner violence (IPV), adverse childhood experiences (ACE) and post-traumatic stress disorder (PTSD); and iii) generation of individual reports with cross-references of statistical analysis between the data obtained in each interview. To validate the tool, a case study was conducted with 50 women assisted in basic health units in a city of the Brazilian Amazon for prenatal care. The results are satisfactory for the use of the tool, which was able to infer emotions (joy, surprise, sadness, and anger), and the prevalence of sadness (25.24%) was verified among the interviewees. For ACE, the majority (21) of the women reported having suffered only physical abuse; as for IPV, the majority of the interviewees (27) reported no abuse; and 78% of the women (39) had no indicative signs of PTSD. The results further point out that there is 3.94 more chance that the group of women who reported any abuse, either in childhood or adulthood, compared to the reference group, would develop PTSD.