Harnessing information from injury narrative in the 'big data' era: Understanding and applying machine learning for injury surveillance

Objective Vast amounts of injury narratives are collected daily and are available electronically in real time and have great potential for use in injury surveillance and evaluation. Machine learning algorithms have been developed to assist in identifying cases and classifying mechanisms leading to injury in a much timelier manner than is possible when relying on manual coding of narratives. The aim of this paper is to describe the background, growth, value, challenges and future directions of machine learning as applied to injury surveillance. Methods This paper reviews key aspects of machine learning using injury narratives, providing a case study to demonstrate an application to an established human-machine learning approach. Results The range of applications and utility of narrative text has increased greatly with advancements in computing techniques over time. Practical and feasible methods exist for semi-automatic classification of injury narratives which are accurate, efficient and meaningful. The human-machine learning approach described in the case study achieved high sensitivity and positive predictive value and reduced the need for human coding to less than one-third of cases in one large occupational injury database. Conclusion The last 20 years have seen a dramatic change in the potential for technological advancements in injury surveillance. Machine learning of ‘big injury narrative data’ opens up many possibilities for expanded sources of data which can provide more comprehensive, ongoing and timely surveillance to inform future injury prevention policy and practice.

Short biceps femoris fascicles and eccentric knee flexor weakness increase the risk of hamstring injury in elite football (soccer): A prospective cohort study

Background/Aim: To investigate the role of eccentric knee flexor strength, between-limb imbalance and biceps femoris long head (BFlh) fascicle length on the risk of a future hamstring strain injury (HSI). Methods: Elite soccer players (n=152) from eight different teams participated. Eccentric knee flexor strength during the Nordic hamstring exercise and BFlh fascicle length were assessed at the beginning of pre-season. The occurrences of a HSI following this were recorded by the team medical staff. Relative risk (RR) was determined for univariate data, and logistic regression was employed for multivariate data. Results: Twenty-seven new HSIs were reported. Eccentric knee flexor strength below 337N (RR = 4.4; 95% CI = 1.1 to 17.5) and BFlh fascicles shorter than 10.56cm (RR = 4.1; 95% CI=1.9 to 8.7) significantly increased the risk of a subsequent HSI. Multivariate logistic regression revealed significant effects when combinations of age, previous history of HSI, eccentric knee flexor strength and BFlh fascicle length were explored. From these analyses the likelihood of a future HSI in older athletes or those with a previous HSI history was reduced if high levels of eccentric knee flexor strength and longer BFlh fascicles were present. Conclusions: The presence of short BFlh fascicles and low levels of eccentric strength in elite soccer players increase the risk of a future HSI. The greater risk of a future HSI in older players or those with a previous HSI is reduced when they possess longer BFlh fascicles and high levels of eccentric strength.

DAQ00152 Herbicide Tolerance Screening in Northern Region (Phase IV)

The previous projects (phase I - III) highlighted that northern region wheat and barley cultivars differ considerably in their sensitivity to herbicides. The new project will focus on increased screening of advanced breeding lines and new cultivars lines to commonly used herbicides, for barley, chickpea and wheat. Studies on impact of environment on herbicide x genotype responses will also be undertaken with the national team. The new information will be added to the existing information package on herbicide tolerance. Thus, adverse impacts of herbicides on productivity in northern region will be reduced, as growers and agronomists will select safer herbicides for their sown variety, or select more tolerant varieties for their important herbicides.

Australian Herbicide Resistance Initiative (phase 4)

A national focus on strategic and applied research to minimise herbicide resistance in Australian cropping.

Innovative management of herbicide resistant weeds - seed banks and soil biology

The project will evaluate seed bank depletion of key northern herbicide resistant weeds under different environments, cropping systems, crop agronomies and non-chemical control tactics. The project will also evaluate soil biology and seed bank relationships to explain differences in seed bank persistence.

Risk assessment and preventive strategies for herbicide resistance in NR (phase III)

The threat and management of glyphosate# resistant weeds are major issues facing northern region growers. At present five weeds are confirmed glyphosate-resistant: barnyard grass, liverseed grass, windmill grass, annual ryegrass and flaxleaf fleabane. This project used 25 experiments to investigate the ecology of the grass weeds, plus new or improved chemical and non-chemical control tactics for them. The refined glyphosate resistance model developed in this project used the experiments' findings to predict the long-term impacts on evolution of resistance and on seed bank numbers of resistant weeds. These data led to revised management and resistance avoidance strategies, which were published in the Reporter newsletter, and via an on-line risk assessment tool. - See more at: http://finalreports.grdc.com.au/UQ00054#sthash.oTkCN4Sk.dpuf

Herbicide tolerance of barley and wheat cultivars in NR (phase III)

During the past 10 years, this project tested 23 barley and 51 wheat varieties with 19 and 34 registered herbicides, respectively. It concentrated on new varieties and herbicides. The research highlighted that Northern Region (NR) wheat and barley varieties differed considerably in their sensitivity to these herbicides. Overall, 9 per cent of wheat variety x herbicide combinations and 6 per cent of barley variety x herbicide combinations had significant yield losses (3 to 38%) from herbicides at recommended rates and crop stages. In addition, 21 to 23 per cent had significant yield losses from herbicides at double rates, indicating a narrow margin of crop safety.

Weed management strategies for farming systems with herbicide tolerant cotton.

The introduction of glyphosate tolerant cotton has significantly improved the flexibility and management of a number of problem weeds in cotton systems. However, reliance on glyphosate poses risks to the industry in term of glyphosate resistance and species shift. The aims of this project were to identify these risks, and determine strategies to prevent and mitigate the potential for resistance evolution. Field surveys identified fleabane as the most common weed now in both irrigated and dryland system. Sowthistle has also increased in prevalence, and bladder ketmia and peachvine remained common. The continued reliance on glyphosate has favoured small seeded, and glyphosate tolerant species. Fleabane is both of these, with populations confirmed resistant in grains systems in Queensland and NSW. When species were assessed for their resistance risk, fleabane, liverseed grass, feathertop Rhodes grass, sowthistle and barnyard grass were determined to have high risk ratings. Management practices were also determined to rely heavily on glyphosate and therefore be high risk in summer fallows, and dryland glyphosate tolerant and conventional cotton. Situations were these high risk species are present in high risk cropping phases need particular attention. The confirmation of a glyphosate resistance barnyard grass population in a dryland glyphosate tolerant cotton system means resistance is now a reality for the cotton industry. However, experiments have shown that resistant populations can be managed with other herbicide options currently available. However, the options for fleabane management in cotton are still limited. Although some selective residual herbicides are showing promise, the majority of fleabane control tactics can only be used in other phases of the cotton rotation. An online glyphosate resistance tool has been developed. This tool allows growers to assess their individual glyphosate resistance risks, and how they can adjust their practices to reduce their risks. It also provides researchers with current information on weed species present and practices used across the industry. This tool will be extremely useful in tailoring future research and extension efforts. Simulations from the expanded glyphosate resistance model have shown that glyphosate resistance can be prevented and managed in glyphosate-tolerant cotton farming systems. However, for strategies to be successful, some effort is required. Simulations have shown the importance of controlling survivors of glyphosate applications, using effective glyphosate alternatives in fallows, and combining several effective glyphosate alternatives in crop, and these are the key to the prevention and management of glyphosate resistance.

Evaluating the efficacy of the EZ-Ject herbicide system in Queensland, Australia

The EZ-Ject herbicide system was evaluated as a stem injection method for controlling woody weeds in a range of situations where traditional chemical application methods have limited scope. The equipment was trialled on three Queensland weed species; pond apple (Annona glabra), velvety tree pear (Opuntia tomentosa) and yellow oleander (Cascabela thevetia); at five different cartridge densities (0, 1, 2, 3 and 4) and with two herbicides (glyphosate and imazapyr). Cartridges filled with imazapyr were significantly more effective at controlling the three woody weed species than those filled with glyphosate. Injecting plants with three imazapyr cartridges resulted in plant kills ranging from 93 to 100%, compared with glyphosate kills of 17 to 100%. Pond apple was the most susceptible species, requiring one imazapyr cartridge or two glyphosate cartridges to kill 97 and 92% of the treated plants. Plant mortality increased as the number of cartridges injected increased. Mortality did not differ significantly for treatments receiving three and four imazapyr cartridges, as these cartridge densities met the criterion of injecting one cartridge per 10-cm basal circumference, a criterion recommended by the manufacturers for treating large plants (>6.35 cm in diameter at breast height). The cost of treating a weed infestation of 1500 plants ha–1 with three cartridges per tree is $1070 ha–1, with labour costs accounting for 16% of the total. The high chemical costs would preclude this technique from broad-scale use, but the method could have application for treating woody weeds in sensitive, high conservation areas.

Testing the efficacy of low volume herbicide applications on Chromolaena odorata

Siam Weed (Chromoleana odorata) is the target of an eradication program in north Queensland; however some infestations occur on ground inaccessible to high volume, ground based herbicide spray equipment. Four foliar herbicides were applied to dense infestations of mature Siam Weed in March 2009, near Townsville, north Queensland. Low volume, high concentration solutions containing 40 g L-1 a.i. glyphosate, 1.2 g L-1 a.i metsulfuron-methyl, 10 g L-1 a.i. fluroxypyr + 0.7 g L-1 a.i. aminopyralid and 15 g L-1 a.i. triclopyr + 5 g L-1 a.i. picloram + 0.4 g L-1 a.i. aminopyralid were applied using a 5 L backpack and hand gun (or splatter gun). Relatively small amounts (approximately 24-28 mL) of the high concentration solutions were applied to each bush and assessments of the replicated treated and untreated control plots were conducted 76, 207 and 356 days after treatment. These assessments demonstrated that the fluroxypyr and triclopyr based herbicides controlled 96 to 100% of plants. The metsulfuron-methyl and glyphosate based herbicides controlled 40 and 57% of plants respectively 12 months after treatment, when 3% of untreated control plants were dead. The trial demonstrated that this application method and either of two herbicides provides an additional tool for controlling Siam weed in remote areas, which are inaccessible to traditional higher volume foliar herbicide applications. Lower volume herbicide solutions reduce the volume of water and thus the effort needed to effectively treat less accessible infestations.

Fuel manipulation with herbicide treatments to reduce fire hazard in young pine (Pinus elliottii × P. caribaea) plantations in south-east Queensland, Australia

Wildfire represents a major risk to pine plantations. This risk is particularly great for young plantations (generally less than 10 m in height) where prescribed fire cannot be used to manipulate fuel biomass, and where flammable grasses are abundant in the understorey. We report results from a replicated field experiment designed to determine the effects of two rates of glyphosate (450 g L–1) application, two extents of application (inter-row only and inter-row and row) with applications being applied once or twice, on understorey fine fuel biomass, fuel structure and composition in south-east Queensland, Australia. Two herbicide applications (~9 months apart) were more effective than a once-off treatment for reducing standing biomass, grass continuity, grass height, percentage grass dry weight and the density of shrubs. In addition, the 6-L ha–1 rate of application was more effective than the 3-L ha–1 rate of application in periodically reducing grass continuity and shrub density in the inter-rows and in reducing standing biomass in the tree rows, and application in the inter-rows and rows significantly reduced shrub density relative to the inter-row-only application. Herbicide treatment in the inter-rows and rows is likely to be useful for managing fuels before prescribed fire in young pine plantations because such treatment minimised tree scorch height during prescribed burns. Further, herbicide treatments had no adverse effects on plantation trees, and in some cases tree growth was enhanced by treatments. However, the effectiveness of herbicide treatments in reducing the risk of tree damage or mortality under wildfire conditions remains untested.

Muscle activation patterns in the Nordic hamstring exercise: Impact of prior strain injury

This study aimed to determine: 1) the spatial patterns of hamstring activation during the Nordic hamstring exercise (NHE); 2) whether previously injured hamstrings display activation deficits during the NHE, and; 3) whether previously injured hamstrings exhibit altered cross-sectional area. Ten healthy, recreationally active males with a history of unilateral hamstring strain injury underwent functional magnetic resonance imaging (fMRI) of their thighs before and after 6 sets of 10 repetitions of the NHE. Transverse (T2) relaxation times of all hamstring muscles (biceps femoris long head, (BFlh); biceps femoris short head (BFsh); semitendinosus (ST); semimembranosus (SM)), were measured at rest and immediately after the NHE and cross-sectional area (CSA) was measured at rest. For the uninjured limb, the ST’s percentage increase in T2 with exercise was 16.8, 15.8 and 20.2% greater than the increases exhibited by the BFlh, BFsh and SM, respectively (p<0.002 for all). Previously injured hamstring muscles (n=10) displayed significantly smaller increases in T2 post-exercise than the homonymous muscles in the uninjured contralateral limb (mean difference -7.2%, p=0.001). No muscles displayed significant between limb differences in CSA. During the NHE, the ST is preferentially activated and previously injured hamstring muscles display chronic activation deficits compared to uninjured contralateral muscles.

Reducing the risk of herbicide runoff in sugarcane farming through controlled traffic and early-banded application

The off-site transport of agricultural chemicals, such as herbicides, into freshwater and marine ecosystems is a world-wide concern. The adoption of farm management practices that minimise herbicide transport in rainfall-runoff is a priority for the Australian sugarcane industry, particularly in the coastal catchments draining into the World Heritage listed Great Barrier Reef (GBR) lagoon. In this study, residual herbicide runoff and infiltration were measured using a rainfall simulator in a replicated trial on a brown Chromosol with 90–100% cane trash blanket cover in the Mackay Whitsunday region, Queensland. Management treatments included conventional 1.5 m spaced sugarcane beds with a single row of sugarcane (CONV) and 2 m spaced, controlled traffic sugarcane beds with dual sugarcane rows (0.8 m apart) (2mCT). The aim was to simulate the first rainfall event after the application of the photosynthesis inhibiting (PSII) herbicides ametryn, atrazine, diuron and hexazinone, by broadcast (100% coverage, on bed and furrow) and banding (50–60% coverage, on bed only) methods. These events included heavy rainfall 1 day after herbicide application, considered a worst case scenario, or rainfall 21 days after application. The 2mCT rows had significantly (P < 0.05) less runoff (38%) and lower peak runoff rates (43%) than CONV rows for a rainfall average of 93 mm at 100 mm h−1 (1:20 yr Average Return Interval). Additionally, final infiltration rates were higher in 2mCT rows than CONV rows, with 72 and 52 mm h−1 respectively. This resulted in load reductions of 60, 55, 47, and 48% for ametryn, atrazine, diuron and hexazinone from 2mCT rows, respectively. Herbicide losses in runoff were also reduced by 32–42% when applications were banded rather than broadcast. When rainfall was experienced 1 day after application, a large percentage of herbicides were washed off the cane trash. However, by day 21, concentrations of herbicide residues on cane trash were lower and more resistant to washoff, resulting in lower losses in runoff. Consequently, ametryn and atrazine event mean concentrations in runoff were approximately 8 fold lower at day 21 compared with day 1, whilst diuron and hexazinone were only 1.6–1.9 fold lower, suggesting longer persistence of these chemicals. Runoff collected at the end of the paddock in natural rainfall events indicated consistent though smaller treatment differences to the rainfall simulation study. Overall, it was the combination of early application, banding and controlled traffic that was most effective in reducing herbicide losses in runoff. Crown copyright © 2012