First report of rapid, non-invasive, and reagent-free detection of malaria through the skin of patients with a beam of infrared light

We describe the first application of the Near-infrared spectroscopy (NIRS) technique to detect 4 Plasmodium falciparum and P. vivax malaria parasites through the skin of malaria positive and 5 negative human subjects. NIRS is a rapid, non-invasive and reagent free technique which 6 involves rapid interaction of a beam of light with a biological sample to produce diagnostic 7 signatures in seconds. We used a handheld, miniaturized spectrometer to shine NIRS light on the 8 ear, arm and finger of P. falciparum (n=7) and P. vivax (n=20) positive people and malaria 9 negative individuals (n=33) in a malaria endemic setting in Brazil. Supervised machine learning 10 algorithms for predicting the presence of malaria were applied to predict malaria infection status 11 in independent individuals (n=12). Separate machine learning algorithms for differentiating P. 12 falciparum from P. vivax infected subjects were developed using spectra from the arm and ear of 13 P. falciparum and P. vivax (n=108) and the resultant model predicted infection in spectra of their 14 fingers (n=54). signatures from 12 patients). We applied the model screening feature to simultaneously fit several machine learning algorithms on the data to allow us to compare and select the best predictive model for infection prediction. Malaria infection status was used as the response factor and the spectral signatures from 900-1700nm, colour skin, age and gender were used as predictors.


28
In 2020, an estimated 241 million malaria-related cases and 627,000 malaria-related deaths were 29 reported by WHO [1]. The Plasmodium parasites which cause the disease are transmitted to 30 people by bites of infected female Anopheles mosquitoes. Among them, the two major 31 Plasmodium parasite species are P. falciparum and P. vivax. In 2020, P. falciparum accounted 32 for 98% of estimated malaria cases globally and 99.7% of the cases in the WHO African region. diagnostic testing to all suspected cases. This is particularly crucial in endemic areas where a 40 majority of the malaria infected population has been reported to be asymptomatic [3] as well as 41 in low malaria transmission settings where the proportion of asymptomatic population among the 42 infected individuals can be as high as 60% [4]. Universal diagnosis is expected to prompt and 43 facilitate the treatment of asymptomatic carriers and limit further community transmission. 44 Optical microscopy, rapid diagnostic tests and molecular tests are the three main diagnostic 45 techniques currently available in malaria endemic regions for malaria diagnosis. Each of these 46 techniques has its own advantages and limitations. Microscopy is the traditional way of detecting 47 malaria parasites in stained thick or thin peripheral blood films using Giemsa, Wrights or Fields 48 stains. Thick film blood films are used to detect the presence of malaria parasite whereas thin 49 blood films are often used to confirm the Plasmodium species [5]. It is the most widely used 50 technique for malaria diagnosis due to its low cost, simplicity and its capacity to detect parasites, 51 differentiate Plasmodium species and estimate the parasite concentration. However, microscopy 52 is technically demanding, time-consuming and requires specialized expertise to accurately 53 identify parasites and differentiate species in samples with low parasitaemia or samples with 54 mixed infections [6][7][8]. As the average microscopist detection limits are estimated to 50-100 55 parasites/µL, the likelihood of underestimating infection rates particularly in low transmission 56 settings or among asymptomatic population where parasitaemia is low has been reported [9].

57
Microscopy is also often unavailable in rural settings where power supply can be problematic 58 [10].

59
Rapid Diagnostic tests (RDTs) are designed to detect malaria antigens in blood by targeting 60 falciparum-specific protein such as histidine-rich protein II (HRP-II) or lactate dehydrogenase 61 (LDH) [11,12]. RDTs are simple, relatively cheap and can be used in remote areas without 62 specialized equipment or need for electricity (Murray et al., 2003). However, like microscopy,

63
RDTs can only reliably detect 50-100 parasites/ µL [11,12] shining a beam of light on a sample for approximately 5 seconds and subsequent collection of a 77 spectral signature. The spectral signature is a reflection of the chemical composition of a sample 78 and can be analyzed using supervised machine learning techniques to determine its diagnostic 79 features. Here, we used a handheld NIRS spectrometer to non-invasively collect spectral    Participants presenting with malaria symptoms were scanned with the NIRvascan spectrometer 118 which was connected to a notebook using Bluetooth. Once connected, the participants arm, ear 119 and finger were one after the other placed directly onto the spectrometer's scan window and 120 spectra was collected by pressing the scan button ( Fig 1A). Two spectra were collected from 121 each body part scanned. A total of 60 patients were scanned and a total of 360 spectra were 122 collected. An example of the average raw spectra collected from each body part for malaria 123 positive and malaria negative individuals is shown in Fig 1B and      and the spectral signatures from 900-1700nm, colour skin, age and gender were used as predictors.

158
The best predictive model selected for differentiating infected from uninfected samples was 159 bootstrap forest with 100 trees. The minimum splits per tree was 10 and the minimum size split 160 was 5.

161
The second supervised learning was used to develop models for differentiating P. vivax from P.  Absorption differences were seen from the raw spectra collected from malaria positive and malaria 183 negative patients. Malaria positive patients were generally seen to have higher absorbance values 184 than malaria negative patients (Fig 1B). There were also clear differences between infected and 185 uninfected patients from the second derivative spectra (Fig 2). Distinct absorption bands involved 186 in the differentiation of malaria positive and malaria negative patients were observed within the1st 187 and 2 nd overtone regions and these bands were present in at least two body parts scanned. They 188 include absorption bands around 1120, 1160, 1337, 1370, 1408, 1626 and 1661 nm (Fig 2). Among 189 them, those that were reduced in malaria infected patients include bands around 1160, 1408, and     The average raw spectra ( Fig 1C) and the second derivative spectra (Fig 4A) of P. falciparum and 237 P. vivax infected patients from the ear, arm and the finger indicates that across the entire spectrum, 238 patients with P. falciparum parasites generally absorbed more light compared to those infected 239 with P. vivax. Differences between the two malaria species were observed at absorption bands 240 around 960, 1171, 1412 and 1530nm. However major absorption differences dominated within the 241 1400-1600nm spectral region (Fig 4B).

242
Using boosted tree supervised machine learning, the model that was developed using spectral 100 percent accurate for differentiating patients into these two groups. When the model was used 245 to predict infection using spectra signatures collected from fingers (i.e. the validation set) that were 246 excluded from the model, only one patient from each group was misclassified (N=27). The 247 sensitivity for detecting P. falciparum and P. vivax was 86% (N=7) and 95% (N=20), respectively 248 ( Fig 4C).  lipids are released into red cells [26][27][28] which are also expected to absorb light at specific 266 wavelengths to generate unique absorption bands (Fig 4). 267 We also tested the hypothesis that P. small scale, this preliminary study provides prior evidence to support the potential use of a cheap, 290 battery-operated, portable infrared spectrometer for this purpose. By removing the need to draw 291 blood, non-invasive diagnosis of malaria by NIRS has the potential to revolutionize our ability to 292 rapidly detect malaria in human populations. Non-invasive scanning will also provide a justifiable, NIRS unit (Fig 1) that was assessed under this study is an off-the-shelf spectrometer that could 308 easily be integrated and scaled up into existing programmatic malaria evaluations. However, 309 further work using a larger sample size from multiple malaria epidemiological settings is required 310 to develop robust predictive models incorporating different demographics including malaria 311 species, parasitaemia level, clinical status, age groups, blood groups, skin color and most 312 importantly mixed/co-infections of malaria or other pathogens such as soil transmitted parasites.

313
Future work should also assess the capacity of NIRS to quantify parasitaemia of infected 314 individuals. Although our results are based on a limited sample size and remain inconclusive at 315 this stage, nevertheless our findings demonstrate the potential of NIRS to non-invasively diagnose 316 malaria in infected people and to differentiate P. falciparum from P. vivax infected individuals.

317
The study also represents initial steps towards development of the first non-invasive, light-based 318 technique for other blood-borne viruses and parasites.