Human in the loop optimization of rocker shoes for offloading plantar pressure in the forefoot.

Abstract

Rocker shoes are often prescribed to reduce pressure in diabetic patients with loss of protective sensation but optimal apex position (AP) and apex angle (AA) vary across individuals. This study investigated whether human-in-the-loop optimization (HILO) can offload multiple high-pressure forefoot zones simultaneously without excessively increasing pressure in low-pressure regions. Ten healthy adults with ≥ 2 forefoot zones exceeding 200 kPa were recruited. Peak plantar pressure was measured using in-shoe sensors, and forefoot zones were classified as "at risk" (AR, ≥200 kPa) or "no risk" (NR, <200 kPa). An evolutionary optimization algorithm was used to find optimal AP and AA to reduce peak pressure in all forefoot regions simultaneously. The HILO optimized shoe was compared with a generic personalization algorithm (Malki) and a control shoe. HILO and Malki shoes significantly reduced the number of AR zones compared to control (p < 0.001 and p = 0.005, respectively). While AP settings were similar (HILO = 62.1 ± 1.7 %; Malki = 63.1 ± 3.0 %), HILO resulted in a wider range of AA (96.9° ± 32.6°) compared to Malki (96.9° ± 8.9°). HILO offloaded all zones that Malki did plus four additional zones across three participants, achieving an 81 % success rate versus 65 % for Malki across participants in this study. These findings demonstrate HILO's potential to individually optimize rocker shoe settings for multi-zone offloading. HILO could find settings that deviate from general guidelines, but deliver similar or better performance. Future work should validate HILO in patients and streamline the protocol to improve clinical feasibility.