Biomechanical Modeling in Prosthetics

Abstract

Fitting a prosthetic socket is, even in 2024, more craft than science. A prosthetist wraps a plaster bandage around the residual limb, lets it harden, pours in gypsum, carves the resulting mould by hand -- adding material here, removing it there -- based on experience, palpation, and educated guesswork about where the underlying bone sits and how the soft tissue will deform under load. The patient tries the socket, reports pain in one spot and looseness in another, and the prosthetist modifies accordingly. Three, four, sometimes five iterations later, the fit is declared acceptable. Not optimal - acceptable. About 35 percent of lower-limb amputees still report socket discomfort as their primary complaint, and nearly a third eventually stop wearing their prosthesis altogether because the socket hurts too much. Finite element analysis could, in principle, replace much of this trial and error. Give a computer the geometry of the residual limb (from a CT scan), the mechanical properties of the tissues (from indentation testing), and the forces involved in walking (from a gait lab), and it can predict exactly where the pressure will be highest and how to reshape the socket to distribute it more evenly. The trouble is that most published FEA models of the residual limb are too crude to be clinically useful -- they treat the entire soft tissue mass as a single homogeneous material with a fixed stiffness, when in reality it is a layered sandwich of skin, fat, fascia, and muscle, each with dramatically different mechanical behaviour. We spent four years (2019-2023) evaluating 180 biomechanical prosthetic modelling programmes across our laboratories in Paris, Madrid, and Rome, covering transtibial amputations (n = 54), transfemoral (n = 44), upper limb (n = 42), osseointegrated implants (n = 22), and running-specific prosthetics (n = 18). We scored each programme on geometric accuracy, tissue material modelling, boundary condition realism, experimental validation, and clinical outcome prediction, combining these into a Biomechanical Prosthetic Modelling Quality Index (BPMQI). The index correlated with actual clinical socket fit improvement at r = +0.84 (AUC = 0.884), and the message was clear: programmes that used patient-specific hyperelastic tissue models, calibrated from the individual patient's own tissue stiffness measurements, produced pressure predictions three times more accurate than those using generic linear elastic assumptions -- and this accuracy translated directly into better-fitting sockets, fewer modification cycles, and happier patients.