Hilab System Device in an Oncological Hospital: A New Clinical Approach for Point of Care CBC Test, Supported by the Internet of Things and Machine Learning

Abstract

The complete blood count (CBC) is a highly requested test that is generally restricted to centralized laboratories, which are limited by high cost, being maintenance-demanding, and requiring costly equipment. The Hilab System (HS) is a small, handheld hematological platform that uses microscopy and chromatography techniques, combined with machine learning (ML) and artificial intelligence (AI), to perform a CBC test. This platform uses ML and AI techniques to add higher accuracy and reliability to the results besides allowing for faster reporting. For clinical and flagging capability evaluation of the handheld device, the study analyzed 550 blood samples of patients from a reference institution for oncological diseases. The clinical analysis encompassed the data comparison between the Hilab System and a conventional hematological analyzer (Sysmex XE-2100) for all CBC analytes. The flagging capability study compared the microscopic findings from the Hilab System and the standard blood smear evaluation method. The study also assessed the sample collection source (venous or capillary) influences. The Pearson correlation, Student t-test, Bland-Altman, and Passing-Bablok plot of analytes were calculated and are shown. Data from both methodologies were similar (p > 0.05; r ≥ 0.9 for most parameters) for all CBC analytes and flagging parameters. Venous and capillary samples did not differ statistically (p > 0.05). The study indicates that the Hilab System provides humanized blood collection associated with fast and accurate data, essential features for patient wellbeing and quick physician decision making.

Keywords: AI hematology; POC hematology; fast cbc test; mobile device; platelets count; rbc count; wbc count.

Figure 1

The Hilab System POCT hematology analyzer. The test operator performs the patient capillary puncture with a lancet. Following this, the operator collects a drop of blood (10 μL) through a disposable pipette and deposits it on a chromatographic strip contained in a plastic capsule. The capsule insertion into the Hilab Flow (chromatography strips analyzer) device concludes the HB quantification process. Next, the operator collects the second drop of the sample (80 μL) and deposits it in the Hilab Lens diluent solutions for blood staining and dilution. Finally, the test operator transfers a drop of each solution into the chambers of the disposable hemocytometer and inserts it into the Hilab Lens device (handheld microscopy). Two (Hilab Flow; Hilab Lens) single-use test kits provide the materials used for CBC Hilab test processing, which accompany isopropyl alcohol swabs and curative. In approximately 30 min, the patient and/or physician receive the report signed by Hilab technical responsible and a laboratory analyst through email or SMS.

Figure 2

The single-use diagnostic kits of the Hilab System and sample preparation process. (A) Components of the Hilab Flow test kits: capsule, capsule cap, blood collection pipette, lancet, isopropyl alcohol swab, and curative. (B) Components of the Hilab Lens test kits: capsule (disposable hemocytometer), blood collection pipettes and blood transfer pipettes, mixing bottle of diluent 1 (WBC count), mixing bottle of diluent 2 (RBC and PLT count), support for mixing bottles, lancet, isopropyl alcohol swab and curative. (C) The test operator collects a drop of blood (10 μL) and deposits it on the chromatographic strip, which promotes the RBC lysis and the HB conversion into methemoglobin. The capsule closing and insertion on the Hilab Flow concludes the process. (D) The test operator collects a drop of blood (40 μL) and adds it to the diluent solution 1 (red circle; dilution factor 1:10), which promotes the RBC lysis and differential WBC dye. The diluent solution 2 (blue circle; dilution factor of 1:180) follows the same procedure, enabling the RBC and PLT visualization. After blood homogenization, the operator individually transfers a drop (10 μL) of solutions 1 and 2 to the hemocytometer chambers and then inserts the hemocytometer into the Hilab Lens.

Figure 3

Passing–Bablok regression plot of the method comparison study between the Hilab System and the Sysmex XE-2100. (A) Clinical protocol for method comparison analysis. (B) Average (μ), standard error (SE), and Student-t test p-values are demonstrated for each CBC analyte (n = 550 patients/group). Red graphs represent the analytes related to red blood cells (RBC, HB, HT, MCV, MCH, MCHC), yellow graphs to platelets (PLT), and blue graphs to white blood cells (WBC, NEU, LIN, MON, EOS/BAS).

Figure 4

Bland–Altman plot of Hilab System results for venous (plus K3EDTA) versus fingerstick blood samples. (A) Clinical protocol for anticoagulant influence analysis and equivalence between venous and capillary blood samples. (B) Graphs demonstrate bias, Student paired t-test p-value, the upper limit of agreement (ULA), and lower limit of agreement (LLA) for each CBC analyte (n = 150 patients/group). Red graphs represent the analytes related to red blood cells (RBC, HB, HT, MCV, MCH, MCHC), yellow graphs to platelets (PLT), and blue graphs to white blood cells (WBC, NEU, LIN, MON, EOS/BAS).

Figure 5

Comparison between Hilab System (left) and standard blood smear methodology (right) for blood cell alterations. Figures show the presence of acanthocytes (Panel (A)), anisocytosis (Panel (B)), dacrocytes (Panel (C)), elliptocytes (Panel (D)), macrocytosis (Panel (E)), microcytosis (Panel (F)), large platelets (Panel (G)), band neutrophils (Panel (H); greater digital amplification for better nucleus visualization), and immature cells (Panel (I); greater digital amplification for better nucleus visualization) in blood samples. Blood smears were stained with May–Grunwald–Giemsa dye, and the images were captured by a Global Optics (NO115T1, Global Trade Technology, São Paulo, Brazil) optical microscope (original magnification ×1000).* Images with greater digital amplification.