预测分析与机器学习优先技术白皮书.docx

UsingDataandAItoGainInsightsinto

1YourSafetyProgram

1

ExecutiveSummary

Thiswhitepaperlooksattheuseofdata,datascienceandartificialintelligence(AI)inEnvironment,HealthandSafety(EHS)applicationsandhowitsdeploymentcanreducetheriskofseriousinjuriesandfatalities(SIFs).Exploringkeyusecases–computervision,naturallanguageprocessing,predictiveandprescriptiveanalyticsengines–wecoverthestreamlined,augmentedandfully-automatedworkflowsenabledbyAI-powereddataanalytics.Weexplorehowsmallorganizationsmaybenefitfromopen-sourceorpay-as-you-gopre-built,modularandflexibletools,whilelarge

enterprisesmayproceedwithlengthyandcostlycustomizeddatacollection,modeltrainingandworldwidedeploymentwithgreatercapabilityandeffectivenesswithinEHS.Advancesindataanalytics,artificialintelligence,machinelearninganduser-friendlygraphicaluserinterfacescanprovidepowerfulinsightstosafetymanagementpersonnelfromanydevice,on-siteorremote,throughcloudcomputing.

KeyFindings:

Datacollectedacrossanindustrialenterpriseinvariousforms(i.e.,writtenreports,forms,images,videoandaudio)canallbeusedbymoderndataanalyticsandAIsystemstoderivepowerfulinsightsanddeliveractionableriskpredictions.

AI-assistedcomputervisionoffersautomatedobjectrecognitionfromimagesandvideosforusesincludingspills,fires,PPEadherenceandsiteinspections.ThetechnologycanbefurthercombinedwithEHSsoftwareandothersafetyworkflows.

NaturallanguageprocessingcanrapidlysummarizewrittenreportsandextractquantitativeinsightsandsentimentstohelpEHSpersonnelperformincidentanalysisandmakequickerdecisions.

Predictiveandprescriptiveanalyticsenginescanuselargedatasetstoreviewpermit-to-workrequests,predicttheriskforfutureincidentsanddeliversuggestedsolutionsbasedonbest-practiceguidelinesandhistoricaldata.

Drawbac