System and method for automatic and real-time design and monitoring of safety measurements

Abstract

A computer-implemented method (100) for improving safety measurements in a working environment. The method may comprise a step of evaluating (110) the safety state of the working environment based on a first set of rules, predicting (120) a future safety state of the environment based on the evaluation and determining (130) a second set of rules based on the predicted future safety state. In addition, a corresponding data-processing device and a corresponding computer program, are disclosed.

Description

[0001] SYSTEM AND METHOD FOR AUTOMATIC AND REAL-TIME DESIGN AND MONITORING OF SAFETY MEASUREMENTS

[0002] Field of the invention

[0003] The present invention relates to a computer-implemented method for improving safety measurements in a working environment as well as a corresponding data processing device and computer program.

[0004] Background

[0005] Traditionally, monitoring of safety measurements and adapting of corresponding safety policies is done manually. Accordingly, the design of such policies is only done punctual and only based on a limited source of information. As a result, these practices achieve only sub-optimal performances in policy design. This is because the basis for the decision making is prone to humanmade errors. Additionally, known policy design techniques rely only past analysis including failure analysis within a working environment based on which a new policy is design. Despite the sub-optimal design of such polices, a further drawback of these techniques is that they are only reactive and not proactive.

[0006] WO 2017 / 156399 Ai discloses a method for evaluating Quality, Health, Safety and Environment (QHSE) data and to identify improvements to HQSE processes and operations. However, the presented approach also represents a reactive approach in which the HQSE processes are merely optimized based on past occasions and thus ignores the problem of proactive policy design and thus fails to provide sufficient improvement of safety measurements.

[0007] Against this background, there is a need for improving safety measurements in working environments. Summary

[0008] The above-mentioned problem is at least partly solved by a computer- implemented method, a data-processing device and a computer program according to aspects of the present disclosure.

[0009] An aspect of the present disclosure refers to a computer-implemented method for improving safety measurements in a working environment. The method may comprise the step of evaluating a safety state of the working environment based on a first set of rules. The method may further comprise predicting a future safety state of the environment based on the evaluation. The method may further comprise determining a second set of rules based on the predicted future safety state.

[0010] In an additional or alternative aspect, the method may further comprise deploying the second set of rules in at least a part of the working environment. In an additional or alternative aspect, the method may further comprise evaluating a safety state in the part of the working environment based on the second set of rules.

[0011] In an additional or alternative aspect, evaluating the safety state may comprise identifying according to the first set of rules at least one safety violation within the working environment.

[0012] In an additional or alternative aspect, the method may further comprise determining an efficacy of the first set of rules based on the evaluation of the safety state. The efficacy of the first set of rules may be set to low if at least one safety violation is identified.

[0013] In an additional or alternative aspect, determining the efficacy of the first set of rules may be further based on information associated with a plurality of workers within the working environment. In an additional or alternative aspect determining the efficacy of the first set of rules may be further based on individual information associated with each worker of the plurality of workers within the working environment. In an additional or alternative aspect determining the efficacy of the first set of rules may be further based on information associated with the working environment. In an additional or alternative aspect determining the efficacy of the first set of rules may be further based on information associated with at least one sensor within the working environment. In an additional or alternative aspect determining the efficacy of the first set of rules may be further based on a company business plan. In an additional or alternative aspect determining the efficacy of the first set of rules maybe further based on financial and / or technology market trends. In an additional or alternative aspect determining the efficacy of the first set of rules may be further based on information about a time associated with the evaluation. In an additional or alternative aspect determining the efficacy of the first set of rules may be further based on any combination of the above mentioned.

[0014] In an additional or alternative aspect, the efficacy may comprise a first average compliance rate during a fist predefined duration before deploying the first set of rules for the working environment. In an additional or alternative aspect, the efficacy may comprise a second average compliance rate during a second predefined duration since deploying the first set of rules for the working environment. In an additional or alternative aspect, the efficacy may comprise a ratio between the first and second average compliance rate. In an additional or alternative aspect, the efficacy may comprise a ratio between a highest compliance rate and a lowest compliance rate during the first predefined duration. In an additional or alternative aspect, the efficacy may comprise an indication of a safety improvement associated with the first set of rules. In an additional or alternative aspect, the efficacy may comprise a difference between the safety state and a reference safety state. In an additional or alternative aspect, the efficacy may comprise any combination of the above mentioned.

[0015] In an additional or alternative aspect, predicting the future safety state of the environment may comprise determining a future point in time for which the future safety state is to be predicted. In an additional or alternative aspect, predicting the future safety state of the environment may comprise forecasting the future safety state for the determined future point in time. Forecasting the future safety state may be done using a regression algorithm.

[0016] In an additional or alternative aspect, the predicted future safety state may comprise a future compliance rate.

[0017] In an additional or alternative aspect, the method may further comprise determining that the future compliance rate is below a predefined compliance threshold. In an additional or alternative aspect, the method may further comprise changing at least one rule of the first set of rules. The at least one rule may be responsible for the future compliance rate being below the predefined compliance threshold.

[0018] In an additional or alternative aspect, the method may further comprise estimating a probability of the future compliance rate. Estimating the probability may be done using a classification algorithm. In an additional or alternative aspect, the method may further comprise determining that the probability is above a predefined probability threshold.

[0019] In an additional or alternative aspect, the method may further comprise determining a time duration required for changing the at least one rule. In an additional or alternative aspect, the method may further comprise determining a time duration required to deploy the resulting second set of rules. In an additional or alternative aspect, the method may further comprise determining a future point in time on which the second set of rules is to be deployed. In an additional or alternative aspect, determining the future point in time on which the second set of rules is to be deployed maybe based on the future point in time for which the future safety state is predicted. In an additional or alternative aspect, determining the future point in time on which the second set of rules is to be deployed may be based on the time duration required for changing the at least one rule. In an additional or alternative aspect, determining the future point in time on which the second set of rules is to be deployed may be based on the duration required to deploy the resulting second set of rules.

[0020] In an additional or alternative aspect, the method of any one of the above- mentioned steps may be executed in real-time. In an additional or alternative aspect, the method of any one of the above-mentioned steps may be executed in a predefined periodicity. In an additional or alternative aspect, the method of any one of the above-mentioned steps may be executed on demand.

[0021] An aspect of the present disclosure refers to a data-processing device comprising means for performing the method of any one of the above-mentioned aspects.

[0022] An aspect of the present disclosure refers to a computer program comprising instructions, which when executed by a computer, cause the computer to perform the method of any one of the above-mentioned aspects. Brief description of the figures

[0023] Various aspects of the present invention are described in more detail in the following by reference to the accompanying figures without the present invention being limited to the embodiments of these figures.

[0024] Fig. la illustrates a general overview of improving safety measurements in a working environment according to embodiments of the present invention.

[0025] Fig. lb illustrates an exemplary overview of improving safety measurements in a working environment according to embodiments of the present invention.

[0026] Fig. 2 illustrates an exemplary overview of a method for improving safety measurements in a working environment according to embodiments of the present invention.

[0027] Detailed description

[0028] In the following, certain aspects of the present invention are described in more detail.

[0029] Fig. la illustrates a general overview loo of improving safety measurements in a working environment according to embodiments of the present invention.

[0030] The overview loo depicts three different phases, namely health-safety- environment (HSE) behavior monitoring no, HSE policy design 120 and policy efficacy monitoring 130. The three phases may be conducted one after another or in parallel.

[0031] In the HSE behavior monitoring no phase, corresponding behavior data within the working environment is monitored and collected. A possible way of implementing such a HSE behavior monitoring and data collecting is described in international application PCT / IB2022 / 060375, which is hereby incorporated by reference. The data may be associated with persons which work within the working environment. The data may comprise HSE compliance rate logs (e.g., manual reports of conducted work which may be in accordance with the corresponding policy or not), a number of alarms (i.e., a number of violations against the policy) etc. In addition or alternatively, the data may comprise a track record of a deployed policy. Deploying a policy as referred to within the present disclosure relates to implementing the policy in the corresponding working environment such that the policy applies to the corresponding working environment. Within the present disclosure, a set of rules may be equated with a policy (i.e., a set of rules may be considered as a synonym for a policy). A policy may be defined by a policy type, a policy description, a procurement time, an implementation time, a response time, a cost of the policy or any combination thereof. The policy type maybe a personal protection equipment (PPE) policy, a behavioral type, an environmental type or any combination thereof. In an example, the policy may be of a PPE policy type and describes that certain PPE (e.g., a helmet, a vest etc.) has to be worn within the working environment. In this example, the procurement time may refer to the time required to procure the PPE. In this example, the implementation time may refer to the procurement time and the additional time required to deploy the policy within the working environment.

[0032] In addition or alternatively, the data may comprise environmental variables. The environment variables may be a set of variables describing the working environment for which the policy is and / or was deployed. A set as used within the present disclosure may comprise one or more units (e.g., variables, rules etc.).

[0033] An environment variable maybe a crew specification (i.e., a description of the crew working within the working environment). The crew specification may comprise a crew number (e.g., an ID which identifiers members of the crew), a shift (e.g., night shift or day shift), a crew supervisor or any combination thereof. It is to be understood that the crew specification may further comprise any information suitable for describing the crew.

[0034] An environment variable maybe a set of individual specifications. An individual specification may describe one individual (e.g., a crew member). The individual specification may comprise ethnicity, seniority, role, business unit of the individual or any combination thereof.

[0035] An environment variable maybe a site specification (i.e., a description of the working environment). The site specification may comprise a type of the working environment (e.g., a vessel, a rig, an office etc.), geographic information of the working environment (e.g., GPS coordinates), weather information (e.g., hot, cold, rainy, cloudy etc.), information about a country in which the working environment is located or any combination thereof. It is to be understood that the site specification may further comprise any information suitable for describing the working environment.

[0036] An environment variable maybe sensor information (e.g., (biometric) information collected from sensors of wearables such as smart glasses or smart devices, cameras or any other type of sensor). In addition or alternatively, the data may comprise business variables (e.g., a business plan of a company of the working environment, market trends, business cycles or any combination thereof).

[0037] In the HSE policy design phase 120, a policy is designed (i.e., determined). The policy design may be determined using a policy recommender system as explained with respect to Fig. 2. In essence, what happens in the policy design phase 120 is that a new policy is determined. The new policy may comprise policy description (i.e., what the policy is about), a communication channel (i.e., how the new policy is communicated), a time of implementation (i.e., when the new policy is to be deployed), a place (i.e., where the new policy is to be deployed, e.g., within the working environment, or only within a certain part of the working environment) or any combination thereof.

[0038] In the policy efficacy monitoring phase 130, efficacy of the policy may be determined. For example, an efficacy score may be determined for the policy. The efficacy may be used as an indication whether the policy is sufficient for the time being (i.e., the policy is effective) or whether the policy is to be adapted (i.e., a new policy is to be determined). Fig. 2 explains a possible implementation of how the efficacy of a policy may be assessed.

[0039] Fig. lb illustrates an exemplary overview of improving safety measurements in a working environment according to embodiments of the present invention. In the example shown, the working environment maybe a marine vessel. Fig. lb depicts on the left side HSE behavior data, which may comprise a life-jacket compliance rate 140a and a number of back-pain complaints 140b. The data maybe captured by an Al-based monitoring system (e.g., computer-vision-based analytics software) as described with respect to Fig. la. It is to be understood that a compliance rate as referred to within this disclosure may relate to a certain type of compliance which again may be determined on an organization, group or individual scale.

[0040] As one can see in this example, the data 140a indicates a decreasing compliance rate when it comes to workers wearing their life-jacket. Meanwhile, the data 140b indicates an increasing number of back-pain complaints. The compliance rate of workers wearing their life jacket may be the result of evaluating the safety state of the working environment (e.g., the marine vessel in this example). Evaluating may be based on the currently deployed policy (i.e., a first set of rules). The compliance rate may for example indicate a ratio of a number of occasions in which a life-vest was worn or was not worn in the working environment and a number of monitored occasions in total. The currently deployed policy (i.e., the first set of rules) may for example comprise a policy ID (e.g., an integer number depending on the amount of previously deployed policies within the working environment; in this example 3) and a description (e.g., wear life-jackets always when vessel is in motion).

[0041] Based on the evaluation, a future safety state of the working environment may be predicted. In the example shown, a future compliance rate maybe predicted based on the available data 140a and it may be determined that the future compliance rate is below a predefined value. The predefined value may represent a minimum compliance rate value for which the safety of the working environment may still be ensured. In addition or alternatively, the future safety state may comprise a future number of back pain complaints predicted based on the available data 140b. The future number of back pain complaints and / or the future compliance rate maybe used for determining whether a new policy (i.e., a second set of rules) is to be designed.

[0042] Fig. ib depicts such a scenario in which both, the future life jacket compliance rate 150a and the future number of back pain complaints 150b reach a certain value, which is indicated by the dotted line, based on which it is determined that a new policy is to be determined. In this example, the new policy may for example comprise a new policy ID (e.g. 4), a description (e.g., wear life-jacket only when vessel is in motion and only within range of im from the vessel railing). The policy may also comprise the communication channel being email, the time of implementation being a certain day in the future and the place being a certain vessel. After deploying of the new policy, the efficacy of the new policy maybe monitored and assessed (e.g., by means of determining an efficacy score). As shown on the right side of Fig. ib, the data indicates an increase of the life- jacket compliance rate 160a and at the same time a decreasing number of back pain complaints 160b over a certain monitoring period. Based on this data, the safety state of the working environment may once again be evaluated. In an example, in which an efficacy score is determined, the score may for example be determined by comparing the life-jacket compliance rate before applying the new policy with the life-jacket compliance rate after applying the new policy. In order to obtain the data (e.g., life-jacket compliance rate 160a and number of back pain complaints i6on) the same HSE monitoring solution as for the data 140a and 140b may be utilized.

[0043] Fig. 2 illustrates an exemplary overview of a method for improving safety measurements in a working environment according to embodiments of the present invention.

[0044] In step 210, a policy efficacy assessment is executed. The policy efficacy assessment maybe part of evaluating a safety state of a working environment based on a first set of rules. Evaluating the safety state may comprise identifying according to the first set of rules at least one safety violation within the working environment. The safety violation may have a negative impact on the safety state of the working environment. The evaluation of the safety state may be used to determine an efficacy of the first set of rules. The efficacy of the first set of rules may be set to low if at least one safety violation is identified. A safety violation maybe identified based on a computer-based analysis reporting (e.g., as outlined in PCT / IB2022 / 060375) and / or human-based reporting Additionally or alternatively, the efficacy may be set to low if the safety state of the working environment is determined to be unsafe. A low safety state may indicate that work carried out within the working environment is below a safety threshold.

[0045] The efficacy of the first set of rules may be determined based on information associated with a plurality of workers within the working environment, individual information associated with each worker of the plurality of workers within the working environment, information associated with the working environment, information associated with at least one sensor within the working environment, a company business plan, marker trends, information about a time associated with the evaluation or any combination thereof. The above-described information based on which the efficacy may be determined, may be collected as outlined with respect to Fig. 1 and the HSE behavior monitoring no. Additionally or alternatively, determining the efficacy may utilize additional information about the first set of rules (e.g., policy type, procurement time, implementation time, response time or required resources) as outlined with respect to the HSE behavior monitoring no.

[0046] An efficacy of the policy (i.e., a set of rules such as the first set of rules) maybe defined as a quantification of historical and / or real-time compliance rate changes. An example for such a quantification maybe an efficacy score. The efficacy or the efficacy score may be defined by a first average compliance rate during a first predefined duration before deploying the first set of rules in the working environment, a second average compliance rate during a second predefined duration since deploying the set of rules in the working environment, a ratio between the first average compliance rate and the second average compliance rate, a ratio between a highest compliance rate and a lowest compliance rate during the first predefined duration, an indication of a safety improvement associated with the first set of rules (e.g., a slope of the compliance rate), a difference between the safety state and a reference safety state (e.g., the reference safety state may indicate an optimal safety state of the working environment and which maybe used for comparison) or any combination thereof. Additionally or alternatively, the efficacy may be defined by a ratio of resources spent for deploying the (first) set of rules and a resulting change of the compliance rate. This way, efficacy of a set of rules may be determined as being low if many resources are spent but only minor increases in compliance rates are determined.

[0047] In step 220, a policy compliance rate forecast is executed. The policy compliance rate forecast may be part of predicting a future safety state of the working environment based on the evaluation according to step 210. Predicting the future safety state of the working environment may comprise determining a future point in time for which the future safety state is to be predicted. The future safety state may comprise or refer to a future compliance rate. The future safety state (e.g., the future compliance rate) maybe forecasted for the determined future point in time. This may be done using a regression algorithm. Accordingly, a regression function may be used to forecast the future safety state using the determined future point in time (i.e., the function outputs a future safety state for a given point in time as input). The future point in time may depend on a required implementation time of one or more rules of the (first) set of rules. For example, the implementation time of the rule associated with the longest implementation time may be taken into account for determining the future point in time. The future safety state may then be forecasted for the future point in time, wherein the future safety state represents a safety state of the working environment at the future point in time without changes made to the first set of rules (i.e., it is predicted how the safety state of the working environment will develop if nothing changes). A compliance threshold maybe defined indicating whether a new policy is required or not. The compliance threshold may be predefined (i.e., a static compliance rate value as threshold value) or may be adaptive (i.e., the compliance rate value used as threshold may vary based on for example the future point in time, based on the safety state or any other factors).

[0048] Additionally or alternatively, a probability of the future safety state (e.g., the future compliance rate) may be estimated using a classification algorithm (e.g., a neural network using a softmax layer or other probability calibration methods applied to classification algorithms). The probability may then be used to determine whether a new policy is required. This may be done by comparing the probability against a probability threshold. The probability threshold maybe predefined (e.g., a static value) or adaptive (e.g., the value of the threshold may vary depending on the future point in time for which the future safety state was forecasted).

[0049] If it is determined that a new policy is required (e.g., if the future compliance rate is below the compliance threshold and / or the probability is below the probability threshold), the method may continue with step 230. This refers to the case that the efficacy of the first set of rules (i.e., the currently deployed policy) is low (i.e., not acceptable).

[0050] If it is determined that no new policy is required (e.g., if the future compliance rate is above or at least not below the compliance threshold or the probability is above or at least not below the probability threshold), the method may terminate and / or be executed again if new HSE monitoring data is available (e.g., as explained with respect to Fig. la, phase 110), which may then be used to execute step 210 again based on the new data. This refers to the case that the efficacy of the first set of rules is high (i.e., acceptable). In step 230, a policy recommender system is executed. The policy recommender system maybe part of determining a second set of rules based on the prediction according to step 220. As part of the recommender system a new set of rules (i.e., a second set of rules) may be determined. The new set of rules may comprise a rule or rules different than the first set of rules. Additionally or alternatively, the new set of rules may comprise one or more additional rules, less rules (i.e., rules being part of the previous set of rules were deleted), one or more changed (i.e., adapted) rules or any combination thereof.

[0051] When removing, adding and / or changing one or more rules, it maybe determined that the one or more rules are responsible for the future safety state (e.g., future compliance rate) and the determined efficacy of the first set of rules as determined in steps 210 and 220. The policy recommender system maybe trained based on a plurality of policies (i.e., a plurality of set of rules) implemented in the past with corresponding efficacy-based adjustments (i.e.., adding, removing and / or changing rules). In order to avoid training of the recommender system on wrong or unreasonable predictions, feedback of users entered via a corresponding user interface may be required. Should the feedback indicate a reasonable prediction (i.e., a recommendation by the recommender system), retraining may be permitted. Should the feedback indicate an unreasonable prediction, retraining may be avoided.

[0052] In addition to determining the second set of rules, a recommended timing for deploying the second set of rules maybe recommended. For this purpose, a time duration required for adding, removing and / or changing the one or more rules and to deploy the resulting second set of rules maybe determined. For example, it may be recommended that two rules of the first set of rules are changed. The resulting second set of rules may then comprise the rules of the first set of rules including the two changed rules. It may then be determined for each rule a time duration required for changing the rule. For example, if a first changing of a first rule suggests a change from equipment A to equipment B, this change may require certain time for procurement of the new equipment B. In another example, if a second changing of a second rule may suggest changing from equipment C to D, wherein equipment D may already be procured, this may require a shorter time. Accordingly, the time required for changing the first rule may be longer than the time required for changing the second rule. As a result, the rule (in this example the second rule) associated with the longest time required for changing the corresponding rule may be taken as reference for recommending the timing for deploying the second set of rules.

[0053] For recommending the timing, a future point in time may be determined. The future point in time may indicate a point in time on which the second set of rules is to be deployed. The future point in time maybe based on the future point in time for which the future safety state was predicted according to step 220. This future point in time for which the future safety state was predicted may also be used for filtering out policy recommendations (e.g., changes of rules) which would require more time than available as defined by the future point in time for which the future safety state was predicted. The future point in time may additionally or alternatively be based on the determined time duration required for changing, adding and / or removing the at least one rule.

[0054] In an example, the future safety state was predicted for a future point in time of t+i, wherein t represents a current point in time (e.g., the point in time in which the method is executed) and i represents a time span. For example, t maybe equal to the current year (e.g., 2023) and i may be equal to 1. 1.e., in this example that within one year (i.e., starting from 2023 + one year = in year 2024) the future safety state of the working environment may reach a certain tolerance threshold beyond which the first set of rules and the corresponding safety state is no longer acceptable (for example the compliance rate being below a compliance threshold). It may then be determined which rule is to be added, removed and / or changed and which time duration is required for the adding, removing and / or changing. In this example, the required time duration may be 6 months. Then the future point in time on which the second set of rules is to be deployed may be determined as t + required time duration. In this example, the future point in time on which the second set of rules is to be deployed may be determined as being within the next 6 months. This way it is ensured that the second set of rules is deployed in time so that the safety state of the working environment does not fall under the certain tolerance threshold.

[0055] Once the second set of rules and optionally also the recommended timing for deploying the second set of rules is determined, the method may continue with step 240.

[0056] In step 240, a policy test and small-scale deployment may be executed. As indicated by the dotted line, step 240 may be considered as an optional step. Deploying the second set of rules in at least a part of the working environment maybe part of the policy-text and small-scale deployment execution. Alternatively or additionally, the second set of rules may be deployed in another working environment or part of the other working environment.

[0057] In this step, a part of the working environment may be identified in which the second set of rules maybe deployed for testing. For identifying, statistical subsampling methods such as, but not limited to, experimental design, stratified sampling, random sampling or randomized controlled experiment methods may be used. Identified the part of the working environment may depend on the type of the policy (i.e., the second set of rules), a magnitude of the second set of rules (e.g., how many rules have been changed, how drastic are the changes, the magnitude of the working environment for example with respect to the amount of workers working within the working environment etc.), required resources for deploying the second set of rules or any combination thereof.

[0058] Once identified, the second set of rules maybe deployed within the identified part of the working environment and the method may continue with step 250. It may also be possible that identified part refers to the entire working environment (e.g., if the deployment requires very less resources or if it is determined that the second set of rules is mandatory to be deployed within the working environment). The policy test may refer to a A / B testing wherein group A is the part of the working environment in which the second set of rules is deployed and group B is a control group (i.e., another part of the working environment in which the second set of rule is not deployed). Comparing these groups maybe done by statistical test comparison techniques such as, but not limited to, Student’s t-test, 2-wayANOVA, 1-way AN OVA etc. This set up may improve the follow up step 250 of policy efficacy assessment.

[0059] In step 250, another policy efficacy assessment may be executed. As indicated by the dotted line, step 250 may be considered as an optional step. The assessment may be used for evaluating a safety state of the working environment in which the second set of rules (i.e., a new policy) is deployed according to step 240. The evaluating may be based on the second set of rules. The policy efficacy assessment may be the same as conducted in step 220 with the difference that now the safety state is evaluated based on the second set of rules. In case the efficacy of the second set of rules is determined to be low the method may terminate and / or be executed again if new HSE monitoring data is available (e.g., as explained with respect to Fig. la, phase no), which may then be used to execute step 210 again based on the new data. Alternatively in case the efficacy of the second set of rules is determined to be low, the method may also continue with the policy recommender system of step 230 to determine a (new) second set of rules which may then potentially achieve a high efficacy. Alternatively in case the efficacy of the second set of rules is determined to be low, the method may also continue with the policy test and small-scale deployment of step 240 to deploy the second set of rules in (another) part of the working environment where the second set of rules may achieve a high efficacy. In case the efficacy of the second set of rules is to be determined to be high, the method may continue with the full-scale deployment of the new policy (i.e., the second set of rules) in step 260.

[0060] In step 260, a policy full-scale deployment may be executed. As indicated by the dotted line, step 260 maybe considered as an optional step. If the second set of rules has a high efficacy as determined in step 250, the second set of rules may be deployed within the entire working environment. This way safety measurements in the working environment are improved. After deployment, the method may terminate or be executed again starting from step 210. It may also be possible that the recommender system of step 230, the efficacy assessment of step 210 and / or the forecast of 220 is retrained based on the successful deployment of the second set of rules allowing for a closed loop and continuous improvement of the disclosed method 200.

[0061] It will be apparent to those skilled in the art that numerous modifications and variations of the described examples and embodiments are possible in light of the above teaching. The disclosed examples and embodiments are presented for purposes of illustration only. Other alternate embodiments may include some or all of the features disclosed herein. Therefore, it is the intent to cover all such modifications and alternate embodiments as may come within the true scope of this invention.

[0062] The aspects according to the present invention may be implemented in terms of a computer program which may be executed on any suitable data processing device comprising means (e.g., a memory and one or more processors operatively coupled to the memory) being configured accordingly. The computer program may be stored as computer-executable instructions on a non-transitory computer-readable medium.

[0063] Embodiments of the present disclosure maybe realized in any of various forms. For example, in some embodiments, the present invention maybe realized as a computer-implemented method, a computer-readable memory medium, or a computer system. The steps described within this disclosure maybe automatically performed.

[0064] In some embodiments, a non-transitory computer-readable memory medium maybe configured so that it stores program instructions and / or data, where the program instructions, if executed by a computer system, cause the computer system to perform a method, e.g., any of the method embodiments described herein, or, any combination of the method embodiments described herein, or, any subset of any of the method embodiments described herein, or, any combination of such subsets.

[0065] In some embodiments, a computing device may be configured to include a processor (or a set of processors) and a memory medium, where the memory medium stores program instructions, where the processor is configured to read and execute the program instructions from the memory medium, where the program instructions are executable to implement any of the various method embodiments described herein (or, any combination of the method embodiments described herein, or, any subset of any of the method embodiments described herein, or, any combination of such subsets). The device may be realized in any of various forms.

[0066] Although specific embodiments have been described above, these embodiments are not intended to limit the scope of the present disclosure, even where only a single embodiment is described with respect to a particular feature. Examples of features provided in the disclosure are intended to be illustrative rather than restrictive unless stated otherwise. The above description is intended to cover such alternatives, modifications, and equivalents as would be apparent to a person skilled in the art having the benefit of this disclosure.

[0067] The scope of the present disclosure includes any feature or combination of features disclosed herein (either explicitly or implicitly), or any generalization thereof, whether or not it mitigates any or all of the problems addressed herein. In particular, with reference to the appended claims, features from dependent claims may be combined with those of the independent claims and features from respective independent claims may be combined in any appropriate manner and not merely in the specific combinations enumerated in the appended claims.

Claims

CLAIMS1. A computer-implemented method (200) for improving safety measurements in a working environment, the method comprising the steps of: evaluating (210) a safety state of the working environment based on a first set of rules; predicting (220) a future safety state of the environment based on the evaluation; and determining (230) a second set of rules based on the predicted future safety state.

2. The method of claim 1 further comprising: deploying the second set of rules in at least a part of the working environment; and evaluating a safety state in the part of the working environment based on the second set of rules.

3. The method of any one of the preceding claims, wherein evaluating the safety state comprises: identifying according to the first set of rules at least one safety violation within the working environment.

4. The method of any one of the preceding claims further comprising: determining an efficacy of the first set of rules based on the evaluation of the safety state.

5. The method of the claims 3 and 4, wherein the efficacy of the first set of rules is set to low if at least one safety violation is identified.

6. The method of claims 4 to 5, wherein determining the efficacy of the first set of rules is further based on at least one of: information associated with a plurality of workers within the working environment; individual information associated with each worker of the plurality of workers within the working environment; information associated with the working environment; information associated with at least one sensor within the working environment; a company business plan; financial and / or technology market trends; and / or information about a time associated with the evaluation.

7. The method of claims 4 to 6, wherein the efficacy comprises at least one of: a first average compliance rate during a fist predefined duration before deploying the first set of rules for the working environment; a second average compliance rate during a second predefined duration since deploying the first set of rules for the working environment; a ratio between the first and second average compliance rate; a ratio between a highest compliance rate and a lowest compliance rate during the first predefined duration; an indication of a safety improvement associated with the first set of rules; and / or a difference between the safety state and a reference safety state.

8. The method of any one of the preceding claims, wherein predicting the future safety state of the environment comprises: determining a future point in time for which the future safety state is to be predicted; and forecasting the future safety state for the determined future point in time, preferably using a regression algorithm.

9. The method of any one of the preceding claims, wherein the predicted future safety state comprises a future compliance rate.

10. The method of the preceding claim 9, further comprising: determining that the future compliance rate is below a predefined compliance threshold; and wherein determining the second set of rules comprises: changing at least one rule of the first set of rules; wherein the at least one rule is responsible for the future compliance rate being below the predefined compliance threshold.

11. The method of the preceding claim 10, wherein the method further comprises: estimating a probability of the future compliance rate, preferably using a classification algorithm; and determining that the probability is above a predefined probability threshold.

12. The method of claim 2, 8 and 10, further comprising:determining a time duration required for changing the at least one rule and to deploy the resulting second set of rules; determining a future point in time on which the second set of rules is to be deployed based on the future point in time for which the future safety state is predicted and the time duration required for changing the at least one rule.

13. The method of any one of the preceding claims, wherein the steps of the method are executed in real-time, in a predefined periodicity or on demand.

14. A data-processing device comprising means for performing the method of any one of the preceding claims 1 to 13.

15. A computer program comprising instructions, which when executed by a computer, cause the computer to perform the method of any one of the preceding claims 1 to 13.

Images