Movella: Deployment of Novel Camera System to Detect and Treatment Injuries for Basic Training

The nature of military duty, especially in training and combat, leads to a disproportionate percentage of injuries to personnel, relative to the general population. For example, paratroopers suffer from injuries to the lower body, notably ankle, knee and hip to a far greater extent than many other military occupations. Injured military personnel who are unable to perform their duties at the desired level of proficiency reduce the overall effectiveness of the unit they are assigned to. While the limitations of some injuries are quite obvious, for example a fractured bone, other injuries may be less so, notably those that are not visible externally, such as an ACL injury or a torn hamstring. Hence, it is important to reliably assess the severity of these injuries by having the afflicted personnel perform physical tests and gauge their proficiency in performing those tests.

Although injuries can manifest themselves in many forms, in the majority of cases musculoskeletal (MSK) impairments are the most common ones that can be observed and assessed. MSK injuries (MSKI) can be caused by sudden incidents or by chronic repeated stresses to the human body. Many of these injures can result in 90 or more days of lost or restricted duty time in addition to the cost of treatment. A majority of these injuries arise from overuse strains, sprains and stress fractures, often to lower extremities such as ankle, feet, knee and hip. Back and shoulder injuries are common as well, mostly related to carrying and lifting activities. These injures affect the range of motion of a person’s limbs, with an adverse impact on their physical attributes.

The overarching objective of this project is to implement a platform that determines the level of injury of a person and assesses their readiness to return to duty. More specifically, this will be achieved in the following manner:

• Obtaining a finite list of the most common MSKI that afflict military personnel, resulting in them needing to take time off for recovery on account of being unable to meet the physical requirements of their duties. For the purpose of this study, the focus will be limited to ankle and knee injuries.

• Identification of specific MSK attributes that have the greatest impact on injuries to military personnel regarding performing their duties to the level that is required – these attributes will typically vary from one injury to the next. For the most part, these attributes would apply to specific parts of the human body and be comprised of a subset of power, speed, endurance, balance, flexibility, agility and coordination. In many cases they would have a correlation with the occupation of the person.

• Designing a set of physical tests that would allow the identified attributes to be measured quantitatively over a period of time, with the trends being recorded. These tests would be designed to be able to be conducted efficiently, to allow a large number of personnel to be assessed per unit of time.

• Collecting kinematic and heart rate data from a sizable number of individuals (100 – 250 recommended), both injury-free and injured, for the purpose of measuring physical attributes with which predictive models can be trained. This would be a combination of intrinsic data (obtained from the tests) and extrinsic data (obtained outside of the tests). Data from injury-free subjects provide an insight into normal values of MSK attributes.

• Training predictive machine learning models with the datasets collected (after labeling) to be able to draw inferences regarding the following:
o Generating scores for the degree of severity of the injury at an aggregate level as well as a per-attribute level.
o Laying out the recovery curves of the individual at an aggregate level as well as a per-attribute level.
o Identifying attributes that have a high correlation with the injury.
o Determining whether the individual is able to return to active duty, and at what level.

• Applying the trained predictive models to new subjects (based on results from their physical tests) and generating reports for each of them comprised of a score for their overall MSK condition, per-attribute scores, and an estimated recovery time when they could return to active duty