SAP Injury Risk Monitor :IoT for Cricket
Recently I got the opportunity to work on a proof of concept solution to help team physiotherapists and doctors of cricket teams to monitor and mitigate the risk of injury to Pace Bowlers.
Who did we work with?
We worked together with Andrew Leipus, physiotherapist of an Indian Premier League team, Dr. Eduard Rene Ferdinands, cricket biomechanics researcher at the University of Sydney, as well as with Dr. Oren Tirosh biomechanics researcher and director at Motion.3D for input and validation of the solution. SAP will be presenting at the 5th World Congress of Science and Medicine in Cricket on March 25 and 26 in Sydney to gain further insights from experts to refine the design and functionality of the proof of concept.
Pace Bowlers in cricket are higher predisposed to injury compared to any other player position in the game. In recent times due to the introduction of the shorter version of the game and also the formation of the Indian Premier league, the calendar is more packed with more games and the pace bowler’s rate of injury has increased dramatically. Few cricket playing nations such Australia, England, South Africa have been recording, monitoring and studying cricketing injuries since the past decade. These surveillance studies have indicated that the reasons for pace bowling injuries are multi factorial. The main factors that put a player predisposed to an injury are workload, technique and the fitness. Workload is the number of balls the player bowls per day and the work that the player did in other player positions like fielding and batting. Whether the player does “over bowl” than the recommended number of balls or “under bowl”, there is a certain amount of risk to injury that is introduced. The monitoring of workload variation is also equally important – for example if the bowler is part a T20 squad bowling just 4 over per match for a month and suddenly called in for “national duty” to play a test match where he needs to probably bowl 20 overs a day. This sudden variation can increase the risk of injury. The technique adopted by each pace bowler is different and is an important factor to determine the likely hood of certain type of injuries. For example a lumbar stress fracture can be largely attributed to an improper technique being adopted. Physical fitness is another good indicator of risk of injury and a close monitoring of this factor is required to know if the player has not been bowling at the near best of his normal abilities. Though these are the main factors there are others as well for example the history of injury, age etc. that does play a role in determining the risk of injury.
While these studies have identified these factors, in order to effectively monitor, calculate and predict the risk of injury is challenging. The collection of data need to be seamless and should be done automatically in an unobtrusive way so that it does not hamper the player ability to perform. Additionally we need a system that can put these factors together to predict the risk of injury and more importantly identify which injury type the player is more predisposed to. There are different kinds of injuries which we refer to as injury types in our proof of concept application and the calculation of the injury risk for each of these types can be different. For example the lumbar stress fracture can be attributed to the technique while the risk of hamstring strain can be attributed more to the workload and poor physical fitness.
Solution and Design:
Team physiotherapists and doctors today make their judgment subjectively and use their experience and skills to determine if a player is fit for playing. A tool that can collectively present all the data in a meaningful way can help them in this decision making process by bringing in more objectivity. So we kept this basic tenet of the design in mind, that, it has to be simple to use and present the data more objectively.
A quick glance of the dashboard of our proof of concept application gives an overview of the predisposal to risk of injury for the players. A risk score gives a good idea of the extent of the risk of injury. Certain players have an orange border around them to indicate that they have been put through some sort of intervention program for risk mitigation. Examples of intervention programs can be for example :- to put the player through a pilates exercise program, regular knee check-up or to rest for a few days etc. A few screen captures of the proof of concept application – the dashboard, player details screen and the mitigate risk screen are shown below.
The way we organized the application is that one could define the concept of an injury type and each injury type can be attributed to several KPIs(key performance indicators). The weightage of the each of the KPI to calculate the risk for the injury type can be changed in the configuration.
The same sensor can be used to collect the player’s sprint speed. The bowling run-up speed can be as low as 60% of the players sprint speed and which can also indicate the potential of the player to increase his effective bowling run-up speed.
First measure the sprint speeds
Set the benchmark speed
Compare the run-up speed for each ball to the sprint speeds
Our proof of concept demonstrates that a simple sensor to measure speed can be used to analyze the varying patterns and check if the pattern is within the normal range and detect in advance if there could be any kind of problem with the fitness.
What we have done with this proof of concept application is connecting the various dots- the research that has gone in to analyze what are the causes for bowling injuries, availability of sophisticated sensors and fast analytic tools that can detect patterns and predict future events.
Availability of highly durable, unobtrusive and accurate sensor technology such as wearable sensor shirt, arms bands, wrist bands, high speed motion capture cameras, sensor soles in shoes, etc. make it a possibility to collect most of the KPI data in an automated way.
Once the data is collected and sent to HANA, different mathematical models are applied to detect patterns and predict the probability of injury.
To a large extent injury prevention is possible if the players are put through appropriate intervention programs, at the first hint of a probable injury. The problem is detecting this early enough and with the latest sensor technology and advanced analytics this is very much a possibility today. This application is not just limited to cricket pace bowlers and can be generally applied to any kind of sports where there is a high rate of injury and a requirement of monitoring.
Thank you note:
I would like to thank the support of my colleagues from the Customer Innovation and Strategic Projects (CI&SP) and the APJ marketing and communications teams at SAP in realizing this PoC- Varun Hariharan, Shailesh Jannu, Vivek Bhanuprakash, Qin He, Nishita Gill, Deviprasad Kapileshwari, Nilus Vincent, Kasem Abotel, Perry Manros, Martin Burger, Jennifer Alejandro, Simon Gomes, Butch Anton, David Chambers, Raj Valame and Puneet Suppal.