Researchers at the 3D laboratory of Masanga Hospital in rural Sierra Leone evaluated the functionality of transtibial prosthetic socket shapes designed with artificial intelligence (AI) in a rural, low-income setting.
Since 2023, local staff at the teaching hospital have independently produced transtibial prostheses using software that was developed in-house. The software has a structured workflow that requires minimal user decisions. It integrates three components: Shining 3D for 3D scanning, 3DMedX for prosthetic socket design with AI shape prediction, and Cura for generating print files with fixed settings. The AI algorithm was trained on expert data.
Masanga Hospital is a 120-bed facility that serves 12,000 patients annually, offers surgical training, supports over 300 nurses in training, and is known for amputation surgery and wound care. In 2022, 75 leg amputations were performed (31 transtibial, 44 transfemoral), and 112 in 2023 (54 transtibial, 58 transfemoral), which likely represents a significant portion of amputations in Sierra Leone, according to the study’s authors.
Thirty-four participants with unilateral transtibial amputations but without a functional prosthesis were enrolled in the study. Hospital workers with limited prosthetic experience autonomously produced the transtibial sockets using the software. The sockets were produced using a hybrid manufacturing technique in which the inner surface socket was 3D printed using 3 mm PETG filament and, after fitting, laminated with fiberglass and resin for definitive use.
After provision of the prosthesis, the researchers conducted baseline and ten-week follow-up assessments to evaluate the participants’ satisfaction, functionality, and quality of life.
Assessments included the adapted Orthotics and Prosthetics User’s Survey (OPUS); activity sensors; the Lower Extremity Functional Status (LEFS); disease-specific Health-Related Quality of Life (D-HRQoL); and the Generic Health-Related Quality of Life using the EQ-5D-3L and the visual analog scale (VAS) score. The participants’ personal goals were also evaluated.
The activity sensors confirmed that all participants used the prosthesis. At follow-up, 27 participants completed the assessments. Median prosthetic wear time was 9.3 hours per day, with a median of 3,978 steps per day (ranging from 1,746 to 6,545 steps), of which 39 percent were with the prosthesis (ranging from 16 percent to 76 percent).
Median scores indicated a significant improvement in LEFS from 45 to 52 and D-HRQoL from 25 to 31; EQ-5D-3L and the VAS score did not significantly change; and 56 percent of the participants achieved their desired activity goals.
The study demonstrated the potential of using AI algorithms to predict the shape of transtibial prosthetic sockets, the authors concluded, giving local prosthetists more independence and providing a standardized approach that can benefit countries with limited prosthesis availability.
The open-access study, “Locally produced AI-designed transtibial prosthetic sockets in rural Sierra Leone–a prospective cohort study,” was published in in the Archives of Physical Medicine and Rehabilitation.
