Technologies for intra-operative ligament balancing using machine learning

1 . A method for training a machine learning algorithm to predict one or more ligament balancing steps in a joint replacement procedure, comprising:

receiving, by a processor, a plurality of input data values comprising one or more surgical parameters associated with the one or more ligament balancing of the joint and one or more real-time sensor data indicative of at least one characteristic of the joint, wherein the plurality of inputs data values are inputted into the machine learning algorithm executing on the processor;

computing, by the processor, using the machine learning algorithm, a pattern indicated by the one or more surgical parameters and the one or more real-time sensor data, wherein the machine learning algorithm is trained on a plurality of input data values from previous joint replacement procedures;

computing, by the machine learning algorithm executing on the processor, a next ligament balancing step based on the pattern, wherein data corresponding to the next ligament balancing step is outputted by the processor;

iteratively repeating computation of a subsequent ligament balancing step, based on the one or more real-time sensor data and the one or more surgical parameters, inputted into the processor, wherein one or more ligament balancing results, for each computation, are outputted by the processor;

assessing, by the processor, a success of the ligament balancing, for each of the one or more ligament balancing results, relative to a target ligament balance state identified by the machine learning algorithm; and

performing, by the processor, an iteration of training of the machine learning algorithm with one or more input data values associated with the target ligament balance state.

2 . The method of claim 1 , wherein the one or more surgical parameters comprise one or more demographic data of the patient, muscularity of the patient, ligament condition of the patient, deterioration levels of various joint structures of the patient, collagen levels of the patient, ligament elasticity of the patient, prior injuries of the patient and patient activities.

3 . The method of claim 1 , wherein the joint replacement procedure is a knee replacement and further wherein the one or more real-time sensor data includes one or more of medial force, lateral force, position of the medial contact point and position of the lateral contact point.

4 . The method of claim 3 , wherein the one or more real-time sensor data further includes gap/displacement measurements and the effect of various resections on gap spacing.

5 . The method of claim 1 , wherein the next ligament balancing step comprises an ordered sequence of ligament balancing steps resulting in the target ligament balance state of the joint.

6 . The method of claim 1 , wherein the joint replacement procedure includes procedures selected from a group consisting of computer simulations of ligament balancing procedures, real-world surgeries performed by surgeons, surgeries performed or assisted by a robotic system and cadaver-based ligament balancing procedures.

7 . The method of claim 1 , wherein the assessment of the success of the ligament balancing is based on one or more of patient feedback, opinions of surgeons, force plates and foot mechanics of the patient and analysis of one or more of electromyography data, muscle-firing patterns, joint stability.

8 . The method of claim 1 , wherein the machine learning algorithm is further trained on surgical preferences of one or more surgeons.

9 . The method of claim 1 , wherein the machine learning algorithm is trained for each ligament individually.

10 . The method of claim 1 , wherein the machine learning algorithm is a neural network trained to recognize features in the real-time sensor data and the surgical parameters and to predict the ligament balancing steps based on the recognized features.

11 . A system for intra-operative ligament balancing of a joint of a patient comprising:

at least one processor;

one or more sensors coupled to the processor;

a machine learning model; and

software, stored on one or more non-transitory machine-readable storage media that, when executed by the one or more processors, cause the system to:

receive, by the at least one processor, one or more surgical parameters associated with the ligament balancing of the joint and input the one or more surgical parameters to the machine learning model;

receive, by the at least one processor, from the one or more sensors, real-time sensor data indicative of at least one or more characteristics of the joint and input the real-time sensor data to the machine learning model;

compute, by the at least one processor, using the machine learning model, a pattern indicated by the one or more surgical parameters and the real-time sensor data, wherein the machine learning model is trained on a plurality of input data values from previous joint replacement procedures;

compute, by the machine learning model, a next ligament balancing step based on the pattern, wherein data corresponding to the next ligament balancing step is outputted by the processor;

iteratively repeat, by the machine learning algorithm, computation of a subsequent ligament balancing step based on the real-time sensor data and the one or more surgical parameters inputted into the at least one processor, wherein one or more ligament balancing results, for each computation, are outputted by the at least one processor;

assess, by the at least one processor, a success of the ligament balancing, for each of the one or more ligament balancing results, relative to a target ligament balance state identified by the machine learning model; and

perform, by the at least one processor, an iteration of training the machine learning model using one or more real-time sensor data and one or more surgical parameters associated with the target ligament balancing state.

12 . The system of claim 11 , wherein:

the one or more sensors comprise a load sensor; and

the real-time sensor data is indicative of a force applied to the load sensor by a bone of the joint.

13 . The system of claim 12 , wherein: the joint is a knee and the bone is a femur; and

the real-time sensor data is indicative of a plurality of forces applied to the load sensor by the femur at a plurality of points between flexion and extension of the knee.

14 . The system of claim 13 , wherein the real-time sensor data includes one or more of medial force, lateral force, position of the medial contact point and position of the lateral contact point.

15 . The system of claim 14 , wherein the real-time sensor data further includes gap/displacement measurements and the effect of various resections on gap spacing.

16 . The system of claim 11 , wherein:

the one or more sensors comprise an optical sensor; and

the real-time sensor data is indicative of a displacement between bones of the joint.

17 . The system of claim 11 , wherein a ligament balancing step includes ligament releases and bony cuts.

18 . The system of claim 11 , further comprising:

one or more a robotic components controlled by the software and configured to perform the next ligament balancing step.

19 . The system of claim 11 , wherein the one or more surgical parameters comprise one or more of demographic data of the patient, muscularity of the patient, ligament condition of the patient, deterioration levels of various joint structures of the patient, collagen levels of the patient, ligament elasticity of the patient, prior injuries of the patient and patient activities.

20 . The system of claim 11 , wherein the machine learning model is a neural network trained to recognize features in the real-time sensor data and the one or more surgical parameters and to predict the subsequent ligament balancing step based on the recognized features.